Tournament complete!
Thanks for participating. Read about the results, or continue betting on Manifold.
💬 Proven correct
Welp, I guess I'm out as far as this competition goes. It was nice knowing you all. Even if I sold I'd be so far behind that I can't win anyway.
guyin2ndplace made S$706!
@PatrickDelaney Wow, you really must be bitter over losing some imaginary internet points to still be whining about this six months later.
@RobertGrosse I would say, "disgusted and contemptuous," ...bitter is too light of a word. And I would say, "literally up at night stressed that I even spent any time at all on this at all," rather than whining...whining is far too light, but yeah, less so about the imaginary points and more so about the outright fraud on the part of Salem Center, Richard H., and being asked to participate in a contest run by a white supremacist troll who posts rants under fake names. But yeah, thanks for your concern, glad to have such a self-aware counterpart that clearly understands fully what happened here.
Can someone explain this to me. Did the Salem Center just decide to do the wrong thing. Why?
@Timothy Because it was a completely troll market run by someone who used to write racist rants online under a fake name the type of person who would have no qualms about just running a contest pretending to be an academic study but really is designed to generate data to show whatever they wanted to show from the onset.
Could Salem provide their rationale for this market resolution? Respectfully, in this case the market stated an alternative source would be found if the OWiD source was unavailable. Many market participants clearly felt a source that was no longer reporting the data in question at the time of the market resolution was by definition not available. For future markets, should we understand that an online data source for quantitative data is available even if it reports data values that are significantly incorrect including zero? This information will help us inform our future betting, so it would be helpful to provide it.
Note...in this case filed in the State of Texas... Plaintiff: ..., Predict It, Inc., ..., Richard Hanania, ..., Josiah Neeley, ... https://dockets.justia.com/docket/texas/txwdce/1:2022cv00909/1188469 ... was published at this blog post here: https://richardhanania.substack.com/p/why-im-suing-the-federal-government - note that Josiah Neeley is a co-plaintiff along with Richard Hanania. Note that Josiah Neeley profited S$1,170 on this clearly wrongly resolved bet. Sketchy AF.
@PatrickDelaney you might want to look google Josiah Neeley and read his bio before going down the collusion accusation road...
@PatrickDelaney Ah come on, this isn't going to convince anyone. Just stick to the (correct) fact that the resolution is obviously wrong via the actual text of the resolution criteria. That's simple, clean and sufficient. This just looks silly and conspiratorial.
@PPPP Yeah I had done that already that doesn't do anything for me. I'm not accusing anyone of anything, I'm just saying it looks suspicious in light of the market being resolved so blatantly wrong.
@PatrickDelaney This made my day, thanks.
"if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023."
Using OWiD, China has reported 5.8M 7-day average rolling cases on December 26th, 2022 which fulfills the condition, "at any point between the opening of the market and February 28th, 2023," therefore this market should re-resolve as YES now.
https://ourworldindata.org/explorers/coronavirus-data-explorer?time=2022-08-11..latest&facet=none&Interval=7-day+rolling+average&Relative+to+Population=false&Color+by+test+positivity=false&country=~CHN&Metric=Confirmed+cases
@PatrickDelaney Go home Patrick, we lost...
@IraklisTsatsoulis In fairness, he is entirely correct.
@MatthewLilley the following 2 statements are in no way contradictory: 1) he is correct regarding the no. of cases (as we have been saying for months now) and that the market should have been resolved as Yes 2) Salem decided to resolve otherwise, and we lost (calls for "re-resolving" are a joke)
@IraklisTsatsoulis If those two things are entirely not contradictory, then (your view on) #2 doesn't really invalidate him saying #1. Perhaps it is a waste of time etc but he's entirely justified imho to keep proclaiming #1 as loudly as he wants until the end of time.
By this logic the market should remain open indefinitely because OWiD may change their data at any time. Perhaps years hence OWiD will go down and Salem will finally resort to an alternative choice.
@MatthewLilley I really cannot see your point here, addressing myself; he may very well be justified, and we are all equally so to comment, critique, and suggest otherwise; I never suggested that he should be censored or something...
I love the @RobertGrosse “proven correct” comment
@JosiahNeeley Yeah, I'm really happy that I was proven incorrect. I still can't believe it actually resolved NO.
@RobertGrosse I can't believe it either
Oh, I'm glad to hear that China had less than 100,000 cases per day of COVID and that their government never reported it either. Whew, that's a relief.
After playing both sides of this market for months my net profit was $-17 🧠
And we have a winner - NO! GG everyone - I am out of here...
Response from OWiD, who we all owe a great debt of gratitude for making this possible.:
> Thanks for getting in touch. I just scrolled through the latest comments on Manifold and can see how complicated this situation is. We won't give any particular opinion on how to handle the resolution of the market, but if it helps, here's more information below.
> The data published by Johns Hopkins University (JHU) currently shows 2,023,904 confirmed cases in China.
The data published by the WHO currently shows a total of 99,030,129 confirmed cases in China.
> The reasons for the discrepancy between the two sources are explained by the JHU team here https://github.com/CSSEGISandData/COVID-19/issues/6576.
Note that we (Our World in Data) do not collect data on confirmed cases ourselves; we've always relied on third-party sources for this data.
> The file you mentioned (https://covid.ourworldindata.org/data/owid-covid-data.csv) is our primary COVID dataset aggregating data from multiple sources. This file still relies on the JHU data for confirmed cases and will do so until 8 March.
> On 8 March, we'll merge our pull request to start relying on WHO data instead, and the entire time series (all the way back to early 2020) will be updated in this file.
> I hope this helps!
So in summary, the data trail goes:
Today, March 8th: OWiD Dashboard > OWiD CSV File -> Johns Hopkins Data -> Note to refer to WHO Data Going Back to December 9th, 2022.
After March 8th, 2023: OWiD Dashboard > OWiD CSV File -> WHO Data going back to January 2020
In either case, we see over 100,000 daily new cases during the week of December 19th, 2022 which resolves the market as YES in either scenario.
So given that, if you review https://github.com/CSSEGISandData/COVID-19/issues/6576 - you will see that Johns Hopkins has said that:
* “Any use of the WHO-China data would require extensive back correction to avoid a +100M daily spike. There is no ‘defensible’ date from which to begin the back correction.”
* Cases and deaths data reported by WHO can be found here: https://covid19.who.int/region/wpro/country/cn. We recommend accessing data for WHO after or around December 9, 2022 when the difference between our time series begins expanding.
So the resolution criteria above states, “Should that source be unavailable, a suitable alternative will be found.”
The first dictionary definition of available at dictionary.com is: https://www.dictionary.com/browse/available
“available - [ uh-vey-luh-buhl ] adjective - suitable or ready for use; of use or service; at hand”
OWiD has acknowledged that their data is not suitable for use of service and in doing so has pointed to the substitute data source shown in the above link to the Github Issue from Johns Hopkins for the interim until March 8th, 2023. That link in turn points to the WHO link, which shows well over >100,000 cases for the week of December 19th, 2022, therefore meaning that this market should resolve as YES immediately.
I understand this is very upsetting to a lot of people on this thread since one might be lead to believe that because the number on this betting market is 5%, which on a surface level suggests that surely due to consensus, the market must resolve as NO. However I will point out that the market maker, Richard Hanania has been very vocal in his distaste for, “bad thinking,” and, “predicting things incorrectly. Here are a few examples:
https://twitter.com/richardhanania/status/1241909570768793601?lang=en
> Tom Nichols, self-appointed expert on experts, misrepresents Tetlock's work, which does show that educated laymen match experts and some predict better. Then says that "experts" as he defines them shouldn't be judged on their predictions anyway. As bad a thinker as you can find
https://twitter.com/RichardHanania/status/1527124727466143745
> Told you. My record of predicting American politics is flawless. I just got a payout today after forgetting that in addition to Oz I had bet on the Trump endorsed guys in PA gov and NC Senate based on my “Trump owns the Republican Party forever” theory.
https://twitter.com/richardhanania/status/1617765690693521408
> Legalize prediction markets now and try to create a world where people this stupid are held accountable for the ways in which they pollute public discourse.
Given these past statements, if Hanania were to choose to resolve this market incorrectly, e.g. to either resolve contrary to the previous criteria, given that there is a 4800% discrepancy between 2M total covid cases in China and 98M covid cases in China, or to specifically choose to resolve in a way that goes with what is popular rather than what is correct, given that this is the second largest market in the contest to date, that would seriously jeopardize his reputation with regards to anything he may say regarding the power of prediction markets in the future, because it essentially shows that they are a purpose-built way to choose winners rather than actually scientifically predict anything. Just imagine the ammo that such a clearly wrong prediction gives any of his detractors going forward for years - it could be very painful for him, whereas the upside to him resolving the market correctly as YES, is that he gets to teach a very important lesson to everyone, that you can’t just go with whatever is popular, you need a certain level of academic scrutiny when making statements and opinions for a policy purpose.
I think this discussion would be quite a bit different if we were on more of a knife edge, say a 5% discrepancy or so, I might just stay quiet on the topic, because there’s an acceptable margin of error - but this is literally a 4800% discrepancy in the data trail which would be an absolute joke for any serious policy researcher to accept and resolve as ground truth.
@PatrickDelaney I ain't reading all that. I'm happy for you tho. Or I'm sorry that happened.
@zubbybadger TD;LR me get zubbybadger money.
@SalemCenter please either pay out or reopen the market... I need my liquidity for elsewhere
@Spencer "Please give me a gold star participation trophy and disregard your stated mission of recruiting a research fellow."
@Spencer Prediction: the market will *not* reopen; forecasting what has already happened is of no interest to anyone.
@Spencer ^Very much this. Do want to bet on other markets plz. The opportunity cost on not resolving this is a pretty large tax rn.
🤷♂️
As @CharlesPaul said here: https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#UGMF2FqIOWGtZk1SnkPa
"Remember folks, win or lose, the important thing is to have fun! and whatever else you can say about it, this market has certainly not been boring."
OWiD must now justify their discrepancy in order for this data source to be interpreted as being, "available."
available - [ uh-vey-luh-buhl ] adjective - suitable or ready for use; of use or service; at hand:
Being that their data has a discrepancy of over 4,800%, according to what their pull request says, since they are claiming to draw off of the WHO data, then this source is not suitable for use under that definition of the word, "available." Available does not only mean, "quickly ready for use," it can also mean, "suitable for use." This data is not suitable for use. Therefore the resolution must default to, "government reports," - https://en.chinacdc.cn/news/latest/202301/W020230126558725888448.pdf unless they can justify on that github thread why their data is defensible in its present form.
@PatrickDelaney I said that when I thought I was going to win. 😏 Well, even if this does go against me, I guess I still have to stand by it now.
In all seriousness, though, I think this was a good learning experience for
everyone involved, including Salem. One of the key lessons we can take from this market (which I will remind everyone is fake and doesn’t matter unless you are actually in the running for the fellowship) is that malformed questions are the bane of prediction markets. The way the question was asked, half of the participants read it as an attempt to estimate the true case count in China, whereas the other half read it as asking about what China would report. Even after Salem issued their clarification, things still spiraled after China semi unofficially reported cases in a way which did not show up in OWID. The thing ended up coming down to a data update, which could just as easily have gone the other way. I honestly think that Salem, and anyone interested in market making, should take note and try to create markets which avoid this type of thing, and, if I should keep my place on the leaderboard and get to submit a paper on my thoughts on the market, I precommit to bring up this specific market as an example of what market makers need to do better.
Incidentally, Patrick, I would like to thank you for making a lot of good points, both intentionally and by example, and for ensuring this market was not boring. And take heart, by next week one of us will have made the other insanely rich in Salem bucks.
@CharlesPaul Thank you Paul, you are a gentleman and a scholar. That being said, I think I may slightly disagree with you and others on this betting forum as to the purpose of a, “prediction market.” I would say that first off, “prediction market,” is a misnomer, it’s actually a, “betting market,” because its predictive function is fairly useless, in the sense of novel, scientific predictions - it’s a weasel word made to make our gambling activity sound more sophisticated than it really is. Secondly, I think the end goal should be to collectively collaborate and come up with a more accurate model of reality, rather than lawyer back-and-fourth about who should win. I think adding a monetary incentive, e.g. the $25,000 prize creates something more akin to the stock market, it gets completely detached from reality and it becomes all about hype and where people think the market will go in the future, rather than the actual ground truth. I think it’s relatively pointless, from a research perspective, (since Salem purports itself to be a policy research institution), to create these kind of zero-knowledge incentives based upon a monetary outcome, rather than a completely imaginary point system, with friendly, collaborative discussion on a topic. Basically it would be a much more efficient use of resources to gamify research, rather than to create a wild stock market that makes no sense in reality, which is what they have done here. Why do I think prediction markets are a misnomer, in the sense of novel, demarcation-based predictions? I have set up markets on the main manifold thread that were for things that were extremely boring and have absolutely no room for interpretation, are validated by a third party, for example, the precise lunar coordinates of where the Chang-E rocket was going to crash into the moon, give or take a particular radius, or the temperature at a particular weather station at a particular point in time. The market participation in these true, “empirical predictive enterprises,” was extremely low. However when you look at other markets which involve things that are very unpredictable yet viral in nature, e.g. accessible to a large number of participants, such as the use of nuclear weapons, they attract hundreds of market participants who end up just arguing back and fourth and their arguments all boil down to, “me like winning.” So in this particular market with regard to China, my initial thought was - due to the exponential scaling law, the R value of the Omicron variant, it was evident as early as December 12th, 2022 that there would be hundreds of millions of cases of COVID in China and there was going to be nothing on earth that would stop China from obfuscating that fact. I did a lot of research, found who was reporting the numbers, followed that back to the source and stuck with my bet. Meanwhile, others on here have followed disconnected-from-reality stock market behavior, trying to play the market swings and follow where the consensus goes, and then shout me down. I would say that this squarely goes against the, “death of expertise” blog post that Richard Hanania posted at the outset of this contest and is in fact not lining up with his or Salem’s overall goals.
@PatrickDelaney You argue as if prediction markets were a novelty, first introduced by Salem here, instead of something with already a decades-old history behind it, and an already significant scientific literature documenting and analyzing both their advantages (which you conveniently choose to ignore) and disadvantages; after blaming Salem for this supposed novelty of theirs, you counter-propose to replace such markets with something of your own invention ("a completely imaginary point system, with friendly, collaborative discussion on a topic"), again failing to acknowledge that this already exists, i.e. platforms like Metaculus, INFER, and Good Judgement Open, which act in a complementary way with other tools (like prediction markets). At the end of the day, instead of acknowledging the value of diversity in such tools, each one with its pros and cons, you propose to kill one and keep only the other, based on the by definition limited scope of your personal experience. And since Salem only "purports" to be a research institute, you fail to even consider the possibility that all this contest here may be part of a research program, aiming to shed some more light on the issue. Again (and rather unsurprisingly), you are simply off-topic.
@IraklisTsatsoulis I'm very puzzled by your comments. if you continue to insist on asking me to not discuss things freely, I would prefer you contact me personally. You obviously have a large problem with my comments, having top commented me for a while now and asked me not to speculate on things, saying that I'm off topic, yet speculating on Salem running a research project yourself at the same time (which I have no problem with). You seem to have some direct personal problem with me. If So, I have reached out to you on Twitter and LinkedIn, you can clear up whatever problems you have with me there if you wish. Otherwise, I'm going to just continue to post whatever I want quite frankly.
@PatrickDelaney I have nothing personal against you, thus I am not going to contact you personally (or reply to any such request of yours); it goes without question that you are of course free to post whatever you like, and others are *equally* free to comment and criticize - and that's all, pretty and simple! Sorry if you have an issue with that, or if you tend to see such freedoms only from your side...
Just putting this here at the top to make sure Salem doesn't miss out on it when resolving this market. https://en.chinacdc.cn/news/latest/202301/W020230126558725888448.pdf
@PatrickDelaney Since they have not resolved it yet, I guess they will wait until the WHO data go all the way through OWiD and displayed there, on March 8; not an unreasonable thing to do, given the circumstances.
@IraklisTsatsoulis possibly. They could just be deliberating. They have a huge audience of fans who they potentially have to piss off by resolving YES, but clearly, as you and I both agree, resolving YES is absolutely disconneced from reality. Tough decision for them.
@PatrickDelaney Eh, you might want to rephrase that second last sentence. :-)
@IraklisTsatsoulis If by "they" you mean Salem, and you are guessing they may wait until data drops on OWID c. March 8, you're essentially guessing what I suggested yesterday (not that my suggestion yesterday was a prediction of what they will do, I have no idea, I was merely opining on what a literal reading of the dates would entail).
@MatthewLilley I am afraid I could not figure out what your exact argument was yesterday (and yes, by "they" I mean Salem - who else I could refer to?); it is implicitly clear that Salem has no control whatsoever on the details of data presentation, and they have fully (and understandably) delegate this detail to the data source (as long as it remains available, that is). And given that the relative orders of magnitude (millions of reported cases for several consecutive days VS a 100K threshold for the market), I fail to see how a distinction of "reported/assigned" dates is even relevant here.
@IraklisTsatsoulis seems like a literal reading of the question would exclude this. According to OWID, on March first, China had not reported 100k+ cases. Even if OWID retroactively updates their data to 100k cases, this is being “reported” after March first, and so does not count. The reporting, and not just the cases, have to happen before 3/1. With that said, I do not blame them for taking their time with this market, as it has landed on a bit of an edge case.
@IraklisTsatsoulis "who else I could refer to?" (I just wanted to double check I understood you).
"I fail to see how a distinction of "reported/assigned" dates is even relevant here."
Suppose on March 8 OWID updates to show data for all of Dec 2022 with millions of cases per day. In my nomenclature (I'm not tied to these words, just using them for convenience) they are assigned to December but are reported in March. Hence the question becomes "When Salem says 'between the opening of the market and February 28, 2023', is a literal reading that the window is of "reported" dates or "assigned" dates. My argument was that if you read it as "reported" dates (e.g. as Charles immediately does above) you quickly run into other issues (e.g. that cases hypothetically reported in Dec 2022 assigned to March 2020 would satisfy the resolution criteria, which surely no-one would argue for in good faith). Which by the contrapositive argues for the alternate reading. (Or more mildly, that the wording is poor).
Or to be even more precise, to match my wording yesterday, the distinction is between "Recorded" dates (when they appear on OWID) and "assigned" dates (the date they pertain to once they appear). I prefer this because a) OWID is not reporting anything, it is merely recording it, b) China / (inc via the WHO) actually *reports* the data, c) the question speaks of China reporting the data, not OWID recording it (i.e. "*China* *reports* a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023").
My meta point here is that the people on here arguing for "No" are premising it on "we must be extremely over-literal and since the OWID site does not give a 404 error, that's still the relevant source". My point is that once you go down the path of being extremely over-literal, you cannot pick and choose, and (unlike many other Salem questions) where the date of assignment and date of data release are explicitly separately specified, this one is *at best* worded sloppily.
@CharlesPaul In the market it doesn't say anywhere owid has to report the numbers by 1st of march. It says "if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023" which China has done. Owid has the data from china and is already made it available. They have also done all the work required to change the graph and just need to press one button to deploy it but the maintainer of the data sources, for some reason, decided to do it on 8. march.
@DavidKochanov All three markets that closed yesterday haven't resolved yet; the most natural explanations are either (1) someone is out sick and hasn't gotten around to pushing resolve yet or (2) Salem is waiting until the oil market closes tonight to avoid a race for people to grab the 3% profits from shifting their money over.
@CharlesPaul @MatthewLilley please notice that the market (title + body + clarification) is an integrated body, and it cannot be read piecemeal; I have made clear where I stand on the interpretation here: https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#RSYPNh5L0SKYmxpuymLV , and I cannot see anything in your arguments that brings forward something essentially new.
@IraklisTsatsoulis Fair enough, but if they are going to take this view then the market should be unfrozen. You can make a fair case that OWID has not updated based on cases which China has already reported by 3/1, but if you make that case then you must consider the market still live, and let it remain live until the resolution is clear. That is, Salem either needs to resolve it now or reopen the market, not wait for new updates or events which have not happened yet, such as an OWID update.
@CharlesPaul that's a fair opinion, another one being that Salem should have resolved this as YES long ago, and, if they finally resolve it as YES now, by not doing so on time they have actually *distorted* the market, by creating plausible expectations towards NO. Please notice that I am only trying to guess, and not justify things; it might very well be that my initial guess above (i.e. that they will wait until March 8) is just plain wrong, since, as correctly pointed out by @SamDittmer, this is not the only market that is closed but not yet resolved.
FWIW, I've seen probably 50+ of these scenarios on Predictit and occasionally Kalshi. When the market has been at 90+% one way, I don't think I've ever seen one resolve in the opposite direction.
@IraklisTsatsoulis I still find your theory plausible, I had money in a market on the Ukraine war, which closed about the same time, and it is already resolved. Maybe it is just those three markets that are harder to resolve? As for the “yes” argument, I think, yes or no, we should agree that trading should be allowed to continue until we know what the verdict will be
@CharlesPaul
> I think, yes or no, we should agree that trading should be allowed to continue until we know what the verdict will be
Sorry, I can really not see the argument here why we should do this; doing so would irrevocably *change* the market, which from the very beginning was supposed to end on March 1. Closing it but not resolving it yet *may* be justifiable - well, I guess it depends both on the resolution and the exact justification...
@IraklisTsatsoulis Ok, I guess I am working on the assumption that markets should not be closed until they are ready to be resolved. Saying “the market closes 3/1” I thought would imply that the “this market will resolve based on the situation at 3/1,” and if the determination cannot be made then, then they should have set the deadline until later. I guess they never explicitly stated this in the rules then, so if they want to wait they can do so.
@CharlesPaul it is not an unreasonable assumption in general; but again, neither is *extending* the open period of a market
@CharlesPaul ... sorry, I meant to say that extending the preset open period of a market is *not* reasonable
@IraklisTsatsoulis @CharlesPaul
Why would reopening the market matter (in the sense of being crucially important) either way?
Regarding being closed, stock markets, for example, close trading at times. (There are future markets and after hours trading to some degree, but the general point holds). A prediction market need not be eternally open to be useful.
Regarding reopening - reopening would be functionally identical to creating a new market with identical resolution criteria (that is literally guaranteed to realize the same way as this one). You could hedge (or double down) exposure on this market by trading on the new one. The price on the new one would indicate the updated value of positions on this market. The existence of the new market would in no way invalidate, change or harm this one.
So either seems fine. Market completeness is good (which would favor reopening), but it's not as though this competition has market completeness as a principle.
The market says that the reporting must happen “at any point between the opening of the market and February 28, 2023.” But it doesn’t say which calendar they are using. According to the Julian calendar, today is only February 15th. Therefore this market can only resolve YES.
@JosiahNeeley Bravo, good sir!
Maybe let's not talk about assassinations in a play money market in a tournament on the internet.
Some poor data engineer doesn’t realize he’s got a whole forum dedicated to him hitting a deadline lol
This contest needs more markets. Is there a market creation/submission portal?
The market maker didn't specify resolution date and the condition "This market resolves as YES if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023. Source for settling the market will be Our World in Data at the link below" will be met after owid update the source (before March 10th) so as it is worded I think this should resolve to yes. Also even without all the source changes OWID is not consistent over time. They very often change data retroactively because many of their data sources have several days of lag.
@DavidKochanov I make no claim about how this will be interpreted, but I'll make a positive claim as follows.
Suppose that today, OWID filled in 50k cases for each day in Nov and Dec. That would be ~3m cases reported "today", assigned to Nov and Dec.
If cases are measured by the day they are assigned to, this is a 50k 7dma throughout (most of) Nov and Dec, and wouldn't be sufficient to resolve "Yes".
However, if cases are measured by the day they are reported, ~3m cases in one day puts the 7dma high enough to instantly resolve "Yes".
I don't think anyone has or would (in good faith) argue that cases should be counted by the day they are reported on the OWID site (indeed no-one even keeps track of the 7dma data this way, we wouldn't necessarily know if the threshold was reached - you'd need to continually check the archives.)
But then, by this logic, a plain literal reading of "a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023" would interpret the dates here also as "dates cases are assigned to", not "dates cases are recorded". There's nothing in that text to indicate "oh, we mean average the cases by day they are assigned to, but set the date window by day they are recorded."
And indeed, there is plenty of precedent in other markets for dates of data referring to the date the data is assigned to, not the date of record (any economic indicator, for example).
One may semi-reasonably object that "but obviously they intended to mean x". Perhaps they did. It is not clear, however, that this is relevant. The words say what they say - and the people saying OWID is the only possible source rather need to rely on a literal reading in other ways.
The other natural complaint would be "but since the time series can be revised in the future, reading the dates as date of case assignment means the market can never be resolved (in particular, "No")." I agree this is an issue. This does not stop the value of the asset from attenuating to $0 or $1 though, so this is not intractable. (It does of course show that a better written question would separately specify the dates cases pertain to and the date resolution would be judged at, like is done with the GDP markets. Alas, this cannot be undone now).
@MatthewLilley and @DavidKochanov Extending the market and offering different accounting systems is getting into assassination markets territory, opening up the door for anyone voting, “NO,” - which is the majority of folks on here, to disrupt the OWiD folks, which is highly unethical. The ground truth of the matter is that both the China CDC and NHI showed well over 100,000 cases per day, which I have discussed at length on this discussion board below. The market maker decided not to resolve even though it was clearly already resolved - which in my opinion shows they are a poor market maker, (by normal base Manifold Markets standards, of course it’s their contest so they can do whatever they want). I think it’s time to buck up and if it happens, take the L if that what the market maker wants. If the market maker / Salem / Richard Hanania wants to hire a researcher, or incentivize the type of researcher who is poor at data analysis and just likes to stare at lines on a screen not question what’s behind those lines, that’s Salem’s prerogative. If we start getting into disrupting OWiD, potentially interrupting OWiD work so that they can continue to publish fiction…I am super against that - this just needs to end.
@PatrickDelaney I think only YESers can win by doing assassination level stuff anyway... I am not complaining that they decided to follow the description word for word or asking that they change the Close date. Just saying that they never specified any condition for NO or a date by which the YES conditions have to be met. This means in the word-for-word world they shouldn't resolve to NO as long as YES resolution is still possible.
@DavidKochanov
> they never specified any condition for NO or a date by which the YES conditions have to be met
I am really not following; the market has a clear and unambiguous ending date, and if the condition is not met by then, it should be resolved as No (future revisions are immaterial) - at least that much is apparent and not debatable. What exactly is your point?
@IraklisTsatsoulis if you mean the close date of the market I don't see how the close date is related to the resolution criteria. Its just a date by which you can bet in the market.
@DavidKochanov "This market resolves as YES if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023. " Save possibly for the exact time zone (which is arguably of little importance), what else is unclear here?
okay but if you take this sentence as it is they should have resolved to YES months ago. Otherwise it just specified the relevant time range in OWID
@DavidKochanov This is a different (valid) argument, and has nothing to do with what you say above. Please try to not clutter things (more than they really are, that is).
@IraklisTsatsoulis "What else is unclear here". As worded, it is *at least* unclear whether the dates in that sentence refer to the date of report, or the date the cases are assigned to. Suppose sometime in the past 6 months China had reported a few extra million cases in 2020, revising their old numbers. No-one would argue that qualifies for "Yes", right? But it should if those dates are the dates of report. Which suggests as literally worded, there is a problem.
My point being that the wording, as literally stated, is sloppy. There's a common sense "reasonable person what do they intended" interpretation of course, but once you use that sort of approach it favors not purely relying on OWID since it's not reporting (recording, really) data but other sources are.
@MatthewLilley OK, I see it now, but it still seems far-fetched to me; and in any case, as many of us have already argued here, "not purely relying on OWID since it's not reporting" has a merit on its own, and it does not actually need "enabler" arguments.
OWiD will not come to the rescue on time: "The changes will be reflected in our datasets and documentation on this repository (as well as on our site) on the 8th of March" - https://github.com/owid/covid-19-data/issues/2785
So, it's all back to the previous debate on rules interpretation; let's see...
@IraklisTsatsoulis Seems like GG
@zubbybadger Well, the keyword is "seems"; just FYI (and since you are rather new here):
https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#RSYPNh5L0SKYmxpuymLV
https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#gAo4Bs4wSjJCTTXrem4v
@IraklisTsatsoulis I own a bunch of YES now so might as well join in: Technically, the question only says that it would resolve as per the link given. As written, it doesn’t specify that it would resolve based on fiddling with the GUI shown on that website, instead of the spreadsheet you get when clicking the link “Download the complete Our World in Data COVID-19 dataset”. That has been updated with the WHO data and shows over 100,000 cases some days.
Clearly, this is ambiguous and should resolve N/A to revert my extremely poor life choices. Kinda contrived, but you can bet on it here:
https://manifold.markets/1941159478/will-that-salem-center-china-covid
@1941159478 Thanks. I will not even pretend to know what a N/A resolution might mean; plus, I have downloaded the underlying CSV file (full data), and I cannot corroborate your observation - what I can see is the data as depicted in the graph, with 0 cases reported for each day after Jan 10, and not a single day with > 100K after the market started.
@IraklisTsatsoulis An N/A resolution would unwind all transactions in this market, so nobody would have earned or lost any money trading here: https://help.manifold.markets/c69e1194903747f2aa300ae396dc0259
I think the new data is there, but only in the owid-covid-data-new.csv file as of yet: https://github.com/owid/covid-19-data/blob/master/public/data/owid-covid-data-new.csv
@1941159478 *except opportunity cost
@1941159478 Betting your entire stack, realizing you're probably going to lose, and then requesting an N/A resolution the day before the market closes is LOL funny. Feel free to let me know if you're joking or if you suggested this a month ago or something. Otherwise, that's loser behavior.
@zubbybadger Hey. Cut with the name calling. It was pretty obvious he was complaining good heartily. Hence the "I own a bunch of YES now so might as well join in:" and the "Clearly, this is ambiguous, and should resolve N/A to revert my extremely poor life choices."
@ConnorPitts I wasn’t name calling. If it’s a joke, I take it back completely. This type of thing happens in prediction markets all the time. Buying $4000 worth of Yes and then at the end saying I had the answer all along and this actually resolves Y or at the very least N/A is something that should be pushed back against.
@zubbybadger *thumbs up*
@zubbybadger Desperate times call for desperate arguments ¯\_(ツ)_/¯
I’m mostly joking, though my betting was partly motivated by the resolution criteria not being totally clear. I think this is probably going to resolve NO in the current situation, but that it’s kind of ambiguous (and FWIW also said so back when I held a bunch of NO https://manifold.markets/AndyMartin/will-china-reach-100000-daily-covid#eRPKYKEw3bwQVjOzmgJn ) Anyway, it’s play money and will resolve how it will resolve I guess.
@1941159478 Great! I retract my previous comment fully and have no doubt you’ll win it back shortly (assuming it does resolve No)
@zubbybadger Thanks! And also @ConnorPitts ofc
Remember folks, win or lose, the important thing is to have fun! and whatever else you can say about it, this market has certainly not been boring.
This *seems* to be the final relevant pull request, opened just 1 hour ago - "Data: Start using WHO as main source", https://github.com/owid/covid-19-data/pull/2792
Any official word on the exact time this market is declared as done? Is it Feb 28th at midnight, for a specific time zone?
@ConnorPitts Mar 1, 2023 06:59 am GMT+2 , by hovering over the end date "Mar 1" in the question header (I guess the displayed datetime depends on where you are based)
As someone observing this insane market, anyone have an explanation on what happened here?
@zubbybadger Our World in Data announced they were going to suspend data reporting. Then they announced they were planning to start it again “ideally by February 27th.” But the market closes on February 28th, which creates a lot of uncertainty.
@JosiahNeeley LOL. The word "ideally" basically decides whether the leaderboard flips or not. Gotta love it
@JosiahNeeley To my knowledge OWiD never announced they were suspending anything. Only question is whether they are relying on JHU's null data or WHO's up to date data
@Spencer Should be interesting! The guy who was in 1st (and I would still consider to be in 1st tbh) decided to let 3K ride here, so this is going to swing this entire thing in a pretty big way!
@zubbybadger is it against the rules for us to ask about y'all at the top's order of magnitude of wealth? Have you guys broken into the 5 figures?
@ZachViglianco No idea if it's against the rules but no advantage in me telling you.
@zubbybadger Given all the ups and downs in this market, quite fun to have it come down to "Will software updates and supporting documentation roll out on time or not"
@zubbybadger Not sure if Josiah misunderstands what's going on or knows something I don't, but what he said sounds incorrect to me. OWiD is an API which used to report from Johns Hopkins. Johns Hopkins announced two weeks ago they are suspending COVID reporting on March 13th after having debated back and fourth whether to incorporate WHO data into their feed - basically, they gave up on reporting COVID overall. OWiD announced that they will switch to WHO reporting, which one would have thought meant the OWiD dashboard would clearly switch over to resolve this bet as a YES, unambiguously. If you look into their Github comments it's fairly clear that they aren't going to take into account the millions of cases reported by the WHO back in December, they are just going to have a hard cut off date for whenever the branch is pushed to master (meaning in March, rather than in December). Please correct me if I'm wrong on this Josiah.
As far as the bet goes, you have a group of people on here who think that OWiD is the ground source of truth for this bet, and you have a group of people who think that it's Government reports from China via WHO, and then you likely have people playing the swings.
What a thrillride, at the end of the day I may even manage to end up back where I started
I am very excited to find out if buying this discounted VIP suite on the Hindenburg was as good of a deal as it seems ...
@ZachViglianco don’t worry, on average half of of us will survive this, a far better deal than the original Hindenberg.
The first commit of the scheduled OWID move from JHU to WHO data has just been merged into 'master': https://github.com/owid/covid-19-data/pull/2783
@IraklisTsatsoulis reading the commit I think this is about over... gg well done
@Spencer gg indeed
ah. meant to reply to iraklis. oh well.
@Spencer Fat lady has not sung yet ;)
(Decides to click on this market for the first time) (Reads comments) (Closes out market immediately)
@zubbybadger this is why you're atop the leaderboard :)
Since folks are likely going to be checking this link a lot - here is the link to the OWiD dashboard itself, pre-filtered for new cases on 7-day average for China. https://ourworldindata.org/explorers/coronavirus-data-explorer?facet=none&Metric=Confirmed+cases&Interval=7-day+rolling+average&Relative+to+Population=true&Color+by+test+positivity=false&country=~CHN
OWID has announced that they will be switching from JHU data to WHO data: https://github.com/owid/covid-19-data/issues/2785
"Our complete dataset should now source case/death data from WHO instead of JHU."
"Ideally, all these tasks should be finished by February 27th."
This question just got interesting again :)
@AsherGabara Thanks for your research and posting that.
Please note that they say:
> Ideally, all these tasks should be finished by February 27th.
My position is still that even if they do not update by Feb 27th, this still resolves YES.
@AsherGabara Who needs entertainment, when you have this?
The meltdown when this still resolves NO will be amusing
@Spencer 😂 I agree.
@Spencer I am glad to see that you do not actually subscribe to the argument that, when the market consensus moves against you, you are obliged to change your position, as you seem to implicitly claim below: https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#bsJYFCCWO3FPAu4spCrF . Still, I am genuinely surprised by your *certainty* here; notice that we don't even need *all* 4 changes listed in the relevant Github issue to have been implemented in order for WHO data to start flowing in, only #1 (which seems to be almost ready: https://github.com/owid/covid-19-data/pull/2783 ) and #4.
@IraklisTsatsoulis Again Iraklis Tsatsoulis , please try to chill out. You changed your bet as well, it's OK, everyone is allowed to change their bet. Spencer is making a joke...there is always a slight possibility that they will go with OWiD. Obviously you and I agree that this would be ridiculous, obviously we have different opinions on why it would potentially happen. It's OK for different people to have different opinions than you. Really..it's going to be OK.
@PatrickDelaney I think many of us would appreciate it if you would try to chill out. Clearly many here find your conspiracies about Salem tiresome. Nonetheless, it seems some of us - eg Iraklis and myself nonetheless agree that YES is the only sensible resolution here. I agree with Iraklis - if they resolve this no, I view this as a indictment on them not my abilities as a forecaster. And you have in the past posted information and links that have provided evidence for YES. But the rest of what you have posted at length, alleging conspiracy and such - honestly I think this has acted to suppress the price here! (If you are playing 4D chess, congratulations, you win; please now stop).
@MatthewLilley fair enough. What made you change your mind if I may ask?
@MathewLilley I have been under the impression that no one cared about my investigation into Salem just as no one listened or cared what I said about China CDC, John's Hopkins, and everything else until there was a shift. Well no one except my sweetheart @IraklisTsatsoulis who copied my link research on WHO and then proceeds to top comment everything I say, (ha ha just kidding, love you Irakis, you my boy). Look, the purpose of a betting market is to collaboratively and research things together. I post links and an argument, then you refute that with evidence and an argument. The net result is that everyone is more informed, in theory. When you provide no outside links and just jumps on people's throat, it dampens that effect. Better is to refute the argument, not the person. I invite you to please, provide more info on Salem or George Hanania that would make me think differently. As I have mentioned here more recently, pragmatically, I don't see them resolving the market contrary to their criteria, as they have not done that in the past. I merely suggest that the entire contest is a way to further their world view. I could get more into the politics of Texas, etc. but basically they are looking for us to be stooges and support this idea that non-experts are better than experts in all things if only free market incentives are applied. Ok, fine...that's ridiculous but doesn't mean they are going to resolve blatantly dishonestly per se. Interestingly having read more of Hanania's work he is of the position that China is not a "real" threat to the US, so one might speculate based upon that, that he would want to lend the China CDC MORE credence, or at least give them a fair shake by pointing to the notion that China did in fact work collaboratively with the WHO on Covid numbers when pressed. https://www.palladiummag.com/2020/12/14/chinas-real-threat-is-to-americas-ruling-ideology/&ved=2ahUKEwjZ-Y3Z2qH9AhVmMDQIHan8DKEQFnoECA8QAQ&usg=AOvVaw03qiBzhW31Y7kmGH4ljr3S
@IraklisTsatsoulis Apologies for any misunderstanding but I would not claim to have very high certainty on the outcome of this one. I more meant *if* this still resolves NO, it will be very amusing. I think this is more likely than not to resolve NO as OWiD is going to switch over to JHU after a certain date, but as far as I know they have given no indication that they are going to backdate previous data.
My belief has always been that this market should resolve by whatever OWiD is showing at 11:59pm on Feb 28, as that is clearly what the rules say. I still think that any interpretation of the rules that says that this market resolves on any other data source is nonsensical, as long as OWiD is still live (which it is). My point in my previous post about market consensus was about market consensus regarding *rule interpretation*. My view on the rule interpretation has not changed at all, but I admit there is more uncertainty now on the actual outcome since OWiD might switch over to a new data source and backdate their data in time to resolve this as YES. I probably put that actually occurring at around 40% chance, which is why I am still holding NO here.
@Spencer Indeed, I dared to imagine you actually meant "if", which does make sense; copy, and all good (our different opinions on rules interpretation are already mutually noted and respected). You do raise an interesting point re backdate data; in the 4th sub-issue in the Github link above, they note "Our complete dataset should now source case/death data from WHO instead of JHU", but I won't claim any high confidence on what exactly they mean. My feeling trying to imagine it from an *implementation* viewpoint, is that it makes much more sense to maintain a single dataset for producing the graphs, so, although they will maintain the JHU dataset for historical reasons, I expect the whole graphed data (past & future) to start relying on WHO; but again, this is just my 2 cents...
@IraklisTsatsoulis You may very well be right... at this point I wish I could exit my position, but the nature of the contest is such that I might as well bank on this resolving NO in hopes I can leapfrog up the leaderboard. Congrats on being on the verge of making the top 5 yourself
@Spencer Appreciate the congrats, but I think they are still premature; although a YES seems much more probable now, the fat lady has yet to sung
All right - was just doing some searching and just found this. I think this unequivocally resolves any data chain of custody topics that I have brought up previously below:
https://en.chinacdc.cn/news/latest/202301/W020230126558725888448.pdf
This is an english language version of the report I have posted earlier from the China CDC, which clearly shows Positive nucleic acid testing what seems to be above 100,000 per day threshold from about the 9th of December through the 12th of January, going up around 7 million cases per day on 22 December, 2023. Per the report:
> PLADs had increased gradually, with the highest number of positive tests (6.94 million) on 22 December
Now, if you want to add antigen testing on top of that, you get an additional peak of around ~400k on top of that 6.94M.
Furthermore, this report was cited in this paper which came out on Feb 3rd here as the first citation [1]: https://www.researchgate.net/profile/Zhanwei-Du/publication/368335849_Estimated_COVID-19_Mortality_in_China_December_16_2022_-_January_19_2023/links/63e34723c002331f725ffaa6/Estimated-COVID-19-Mortality-in-China-December-16-2022-January-19-2023.pdf
Keep in mind, that Salem Center is part of UT Austin, much like the COVID-19 Modeling Consortium at UT Austin, where the authors of that study which cites the China report are also from.
So basically we have: 1) New York Times is not valid because it's pointing to Johns Hopkins which points to WHO, their graphic has not been updated. 2) WHO might be questionable due to possible chain of custody or defensibility concerns, but does meet the threshold if you accept China NHC as being equivalent to WHO. 3) From UT's own COVID-19 modeling consortium, also at UT-Austin, they reference the China CDC report which does indeed point to multiple references points well above the threshold in terms of nucleic acid testing.
Looks like a smoking gun to me; you should have tagged @SalemCenter (and leave aside the largely irrelevant paper UT Austin stuff).
@PatrickDelaney Seems pretty clear that they're going to follow the rules and go off of OWiD. Better to sell while you still can
@SpencerHenderson How is it clear?
@PatrickDelaney If Salem believed the rationale you've attempted to provide for why this should be YES, they would have already resolved it
@SpencerHenderson although this is not an unreasonable argument, I think it is far from certain (or "clear"); you may find my speculation below useful as a possible justification of a late Yes resolution: https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#JWpW8I9FrcyWcAHoyFn3 . And I do argue that it is not an unreasonable one, either. ~20 days from now, we may have many "experts" and know-it-alls that will argue "of course it would be resolved as Yes, with the OWID having stopped and the Chinese officially providing data toward yes; wasn't that obvious?"
@IraklisTsatsoulis The point of the contest is to let the best forecaster win. Leaving a market open just to collect more comments cuts against that quite strongly, and is inconsistent with how they have resolved every other market - i.e. very quickly.
We already seem to have some "know it alls" arguing exactly what you describe, in fact that seems to be the biggest volume of comments, even as the market consensus moves against them.
@Spencer as I already said, your initial point is not unreasonable; I am just saying it is neither certain nor clear yet, until the fat lady sings , i.e. Salem resolves the question. What is currently unambiguously clear is 1) we do have China officially reporting more than 100K/cases per 7-day average 2) these data are publicly available 3) OWID has been ignoring these official data and stopped reporting any cases whatsoever and 4) the majority of market participants seem to understand "source unavailable" only as "404 page not found". As a matter of principle, myself I refuse, under these circumstances (including the question rules, as explicitly written and clarified above), to forecast NO here, and *if* Salem turns out to disagree with this, then I make no claim to be a "best forecaster" according to their own definition; they can keep their 25K contract - I will not compromise my rationality principles. Last but not least, market consensus can be (and has been) wrong quite often, as you can probably recall from our discussions in the recent past in another forum (I made quite a hefty amount from the China GDP question here, irrationally priced until the very moment of announcement: https://salemcenter.manifold.markets/SalemCenter/china-gdp-growth-4-or-more-in-2022 ). So, let's just wait the resolution...
@Spencer
"Leaving a market open just to collect more comments cuts against that quite strongly"
BTW, this is not what I meant - I meant *betting* behavior (although rationales make up for valuable material too, I guess)
@Spencer really this is not evidence, this is conjecture which has been stated by Johnny 10 Numbers - who you will notice is no longer the top trader. This is a highly risky strategy. If you look at CSPI's past resolutions there is no real evidence that they blatantly resolve things contradicting their own resolution criteria thus far. https://salemcenter.manifold.markets/home?s=close-date
While I agree with you that they are a highly biased market resolver, I can't find a reason why they would want to actively risk their reputation as a research institution in favor of one question on one tiny contest. As far as I can tell they give out tons of grants per year and this is just one of many research grants. Why would they care to manipulate the market *that* much?
@PatrickDelaney
> I agree with you that they are a highly biased market resolver
I regret to say you seem unable to make a proper comment, i.e. without blurring things and/or agreeing/disagreeing with non-statements: Spencer has never argued that they are "a highly biased market resolver", neither he seems to have implied anything like that; as for their reputation, you have spent an awful lot of words here arguing that they ("this "Research Lab," as they call themselves", according to your own description) actually only care about their strong and extreme biases and convictions, and little else. Some consistency, please...
@IraklisTsatsoulis ... Belief that Salem will resolve contrary to facts of the matter, e.g. that they will resolve according to OWiD because they have not already resolved is in fact, a belief not supported by any evidence, but rather pure conjecture, and assumes some underlying *other* reasoning for why they would resolve a marketplace incorrectly, which may or may not be due to bias, but bias assumes the most, "well-intentioned," reasoning on the part of the market maker. I do not think anyone here sees Salem as actively malicious, so the logical default is that they would hypothetically be biased. Belief that Salem is going to resolve in line with their previous resolutions, which is to say, in line with the resolution criteria, is a pragmatic narrative approach - it assumes that future resolutions will be similar to past resolutions. As far as Salem's reputation as a research lab - it is VERY clear that they do have strong biases, they make it VERY clear and public that they subscribe to a completely ridiculous philosophical school and they plaster it all over their website. I'm very puzzled why you see these two points as contradictory as I am not aware of how being an Ayn Rand / objectivist makes you incapable of being able to abide by the clause:
> This market was never meant to estimate the true number of cases in China, but was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used. If at some point China stops reporting data completely, this market will resolve as NO as long as there was never 100,000 cases reported for any 7-day rolling average.
@IraklisTsatsoulis > actually only care about their strong and extreme biases and convictions, and little else.
Per your quote here:
https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#l8mNfWZLON9z53hk0ulI
> Based upon standard practices, you (and all of us) are supposed to provide references when quoting what *others* say.
Please point to me where I said, "only care about their strong and extreme biases and convictions, and little else." I do not recall saying that.
@PatrickDelaney
> Please point to me where I said, "only care about their strong and extreme biases and convictions, and little else." I do not recall saying that.
Really now?
1. https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#kxefP2aowXOyQGGdX83F
> I believe this is a mistake on the part of the New York Times, but I don't believe that Richard is going to see this as a mistake because he has an extreme bias.
> it's reasonable to believe that Hania [sic] would default to the NYT because he's much more interested in proving his own points, has an axe to grind so to speak, rather than the gentle art of back-and-fourth thinking.
> he lives on Twitter and looking to make a name for himself. He's not an academic interested in studying things, he's a ridiculous pundit with a bunch of money from Texas billionares,
2. https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#FbNYf2jDuYpRXUBEBWed
> I further speculate that the outcome of this study is pre-determined, as this "Research Lab," as they call themselves, subscribes to the, "objectivist," Ayn Rand philosophical school of thought
I trust they are enough... Now, you please tel me where Spencer has said that "they are a highly biased market resolver", with which you rush to "agree"...
> I suspect/speculate Hania [sic] is going to resolve this as NO, in part because he hasn't already resolved it or provided any update.
https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#kxefP2aowXOyQGGdX83F
i.e. exactly what Spencer (and implicitly others, judging by their bets) has argued above, which you again rush to summarily dismiss as a conjecture and not evidence, and a "very risky strategy"...
Swallow memory...
@IraklisTsatsoulis Nope, you misread me. Never said it, you're misreading what I said above and putting words in my mouth. I suspect you just don't comprehend what I'm saying because you want your YES bet to win so badly...no worries. I already explained my rationale on Spencer...please re-read what I wrote. Please try to chill out.
@PatrickDelaney
> I suspect you just don't comprehend what I'm saying because you want your YES bet to win so badly
Yeah, sure! :D
You are free to suspect what you like, but that is again laughable; anyway, the quotes are there, the comments are there, let's leave if I am misreading you or not to the judgement of the interested reader, shall we?
@IraklisTsatsoulis Ah ok, sure. Yes, agreed let's allow the inquiring reader to pour over two extreme nerds fighting on a betting platform put on by Ayn Rand fanboys. I am sure there will be millions of readers. Again, chill out. You are going to be fine.
I've been reading up on WHO governance and trying to figure out how to establish a better chain of custody to really solidify the argument that the WHO reports are equal to the Chinese NHC reports, other than the vague notion of meetings.
One thing I can find, which points to a chain of custody, is this article:
https://www.globaltimes.cn/page/202212/1282989.shtml
This article represents an endorsement of the meeting between the WHO and China's National Health Commission (NHC), and it was published by the English-language Chinese newspaper under the People's Daily, a state-run newspaper. Within that is stated:
> The two sides communicated on the current state of the epidemic, medical treatment, acute care and continued vaccination. The two sides agreed to maintain technical exchanges to facilitate early end of the pandemic.
However, while Johns Hopkins has also been a reliable source and arguably does have a chain of custody while the WHO's chain of custody is more dubious from what I can tell right now, there is a huge problem with the Johns Hopkins data as presented, which can clearly be seen here: https://coronavirus.jhu.edu/region/china
Basically, the public raw Johns Hopkins feed did in fact report the number of deaths on January 18th, 2023 as having a 7-day average of >10,000, which of course came from Chinese CDC reports as we have discussed below. However, on that same dashboard they report no more than ~28k cases on a 7-day average in the couple weeks prior.
Being that the death rate of the Omicron variant has been cited to be around 0.3%, 10k daily average deaths would imply 3.3M daily average cases at some point in the weeks preceding, which is of course 3 orders of magnitude off of the 28k daily average cases figure.
So using the Johns Hopkins data as a source of data with a clear chain of custody, and knowing that they ceased updating case data because they can't reconcile a defensible start date and have themselves pointed toward the WHO as the most reliable current source...we can look at the WHO data:
https://covid19.who.int/region/wpro/country/cn
The WHO reports 11k deaths during the same week that Johns Hopkins reports 10k daily average, so basically we're missing around 60,000 deaths for a particular week on the WHO report.
In summary, we have conflicting reports from the Chinese CDC and the Chinese NHC, if you accept that the WHO represents the NHC numbers. The WHO/NHC numbers definitely seem more, "consistent with itself," which is why I would suspect Johns Hopkins just gave up on using China CDC numbers and pointed toward the WHO (in addition to them closing their program overall).
I suspect/speculate Hania is going to resolve this as NO, in part because he hasn't already resolved it or provided any update. Here's my rationale:
1. He loves the New York Times and gushes over it on Twitter: https://twitter.com/RichardHanania/status/1624815763369635841?cxt=HHwWgsDRkYDVwIwtAAAA
2. Here's another example, more relevant to China/Covid: https://twitter.com/RichardHanania/status/1622357411867815937?cxt=HHwWgsDTkY7e4oMtAAAA
3. In this New York Times interactive report: https://www.nytimes.com/interactive/2021/world/china-covid-cases.html
They refer to the report from the Chinese CDC on January 12th containing 1.27M new cases as being an, "anomaly," and they also refer to this as, "one month's worth of deaths," under data notes.
> More about reporting anomalies or changes Jan. 12, 2023: China reported more than a month's worth of hospitalized new cases and deaths.
So if we are to believe this is one month's worth of data, that would put the maximum average at 42.3k/day.
I believe this is a mistake on the part of the New York Times, but I don't believe that Richard is going to see this as a mistake because he has an extreme bias. The report they are referring to is of course this one: https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_11809/202301/t20230109_263283.html
Which indicates nothing about the new cases reported being one month's worth of data, rather it is contextualized against several other facts not mentioned by the New York Times, which I have gone through previously, namely that the disease was recategorized as a category B disease, that there were in fact ongoing reports that were severely undercounted, and that this report from the Chinese CDC according to the Johns Hopkins discussion on Github, prior to deciding to cease reporting, essentially before they decided to throw everything out and point toward the WHO, was considered new cases on 1/12/2023 for the purposes of discussion. https://github.com/CSSEGISandData/COVID-19/issues/6414
So essentially, the New York Times graphic is excellent, succinct, great reporting, and it's behind the brand name of the New York Times, but it's missing some pieces, so that, "one month's worth of reporting," comes from thin air.
That being said, Hania loves the New York Times, and being that his center subscribes to the Ayn Rand Objectivist School of thought, he has an extreme bias which will point him toward the belief that by definition entrepreneurs and businesses are always better sources of truth than governmental bodies.
So when betting on these markets, you have to consider who the market maker is and what their biases are. This is arguably a complicated data science problem because there are two sources of truth, the Chinese CDC and the Chinese NHC, and there is arguably a lack of data provenance/lineage from the WHO/China NHC, it's reasonable to believe that Hania would default to the NYT because he's much more interested in proving his own points, has an axe to grind so to speak, rather than the gentle art of back-and-fourth thinking.
You just look at his Twitter feed--all he does all day long is just attack people, he lives on Twitter and looking to make a name for himself. He's not an academic interested in studying things, he's a ridiculous pundit with a bunch of money from Texas billionares, as I have mentioned previously.
@PatrickDelaney laughable argumentation (no surprise on that), now on top of personal attacks to someone whose name you do not even bother to spell right. This all thing is just a game, which nobody forced you to play. Please stop, and spare us the toxicity.
@IraklisTsatsoulis fair enough, perhaps I have said enough on Salem and made my point and apologize for any potential toxicity and bad feelings I may have caused you.
@PatrickDelaney I have my own suspicions for how this market will be resolved. I, and this is just a wild guess, think they will look at the biggest number on the OWID 7 day rolling for China, and if it is bigger than 100k, they will resolve yes, else they will resolve it no.
@PatrickDelaney you have made your point indeed, arguing that they will prefer NYT by default, since, as Ayn Rand fans, they have an extreme bias that "entrepreneurs and businesses are always better sources of truth than governmental bodies", conveniently ignoring the explicit clarification that the market "was always to be based on government reports". Now, who is the biased one here, is arguably not hard for anyone to see, judging no less by the length of the mental gymnastics and pseudo-rationalizations they are willing to undertake...
In the spirit of good sportsmanship, I would like to make a quick post playing devil's advocate on the, "NO" side. I think it would be more productive and interesting to me personally to have this type of conversation and have people on the NO side making some solid arguments and doing further research rather than the just constant insistence that OWiD is still a valid reference - which is ridiculous.
OK, so a more solid, "NO" argument would go something like this:
We can't know that China in fact had over 100,000 cases because of lack of data providence. So basically we have to ask...what is the purpose of data analysis in the first place? It's to make decision and analyze risk.
So the advantage of the Johns Hopkins data was of course that all of the data was, "public," based upon public reports, from the Chinese government.
They have now said that we should all refer to the WHO data. So from what I have read so far, which - I would love to be proven wrong in this...because obviously, it would benefit my bet...the WHO talks directly to the China NHC (as opposed to the open reports of the China CDC). So unless there is some kind of public record of this conversation, for all we know...the China NHC is quietly just telling the WHO, "some kind of number X," and we don't know how that number came to be, because it's not public, and then the WHO is, "somehow publishing whatever other number Y." Now it's extremely possible and in fact almost certain in my mind that X=Y, but...in all fairness, I cannot figure out how to be sure of that.
So, this would default back to perhaps opening the door for Salem and/or NO betters to be able to claim, "NHC/WHO data is not valid from the standpoint of data defensibility." What I mean is that, from the standpoint of the China NHC and China CDC, we already know that this data is not defensible in that it doesn't represent the true number of cases in China, for various reasons. But now there's another layer, which is that, from what I can tell so far, we can't even tell what the China NHC said, we can just tell what the WHO said that the China NHC said.
Does that make sense?
Of course the risk of having this conversation for anyone betting, "NO" is that, if we do find that the WHO does publicly post this information, and we're just not aware of how that works yet, then accepting the WHO-China NHC premise in the first place is, "damaging to their claim."
Looking forward to hopefully more productive and interesting discussion on this topic!
Appendix: here's the main document from the WHO that I could find thus far relating to where they got their data from, it was from, "meetings..." from what I can tell: https://www.who.int/news/item/30-12-2022-who-meets-with-chinese-officials-on-current-covid-19-situation
Further...please note that I am not trying to take the market, I am currently 100% all in on this market with zero reserve cash.
@PatrickDelaney Now, who would bother to claim that "this data is not defensible in that it doesn't represent the true number of cases in China,", while the clarification makes crystal clear that "This market was never meant to estimate the true number of cases in China, but was always to be based on government reports" is anyone's guess. There is not any requirement for the initial data (i.e. the ones provided officially to WHO) to be public, as long as they are official.
If you are looking forward to read something interesting, kindly consider reading a book instead of imagining things (like others basing their bets on the expectation that the "conservative" Salem will not believe the Chinese by default!), and inviting us to discuss them (and don't worry, we know you are not trying to take the market).
I have started suspecting that Salem *may* have chosen not to resolve the market as Yes yet in order to see how our betting behavior evolves - it is certainly a *very* interesting experiment so far...
@IraklisTsatsoulis Thanks for your contribution.
@IraklisTsatsoulis Tell me more about why you think Salem may have chosen not to resolve the market in order to see how our betting behavior evolves and how you developed this statement.
@PatrickDelaney it's not a statement, just a speculation (and I will not even insist too much on it); but if I were doing research on evidence-based decision making (as they do, according to their mission statement), and had in front of me such a situation, i.e. 100K cases/day unambiguously from official data since December, I would *die* to find subjects insisting that this is a "No", and to study their pseudo-rationalizations and motivated reasoning. Such stuff is goldmine, and one would not be in a hurry to terminate the experiment (or the data-gathering mechanism) prematurely - they can always justify why they waited until March 1 to resolve this as a Yes...
@IraklisTsatsoulis I think you may be on to something. Reviewing Richard Hanania's Google Scholar profile, he does investigate partisan and other attitudes toward political decision making. https://scholar.google.com/scholar?hl=en&as_sdt=0%2C24&q=richard+hanania&btnG=&oq=richard+ha
Then looking at the original contest launch:
https://www.cspicenter.com/p/introducing-the-salemcspi-forecasting
You can see he links to this form:
https://docs.google.com/forms/d/e/1FAIpQLSfnjpbr6AXHOlYFRWsJi1G8tmiv4yX5Mt1GibkKVocvsbpcjw/
Which includes tons of demographics questions, which remind me of the types of demographics questions submitted to me by a Carlson School of Management Professor at the University of Minnesota that he wanted my conference attendees to fill out for a conference I used to run.
So basically, the speculation by my understanding is that, "We could be in a large experiment trying to figure out how biases affect performance in betting on real world outcomes."
I further speculate that the outcome of this study is pre-determined, as this "Research Lab," as they call themselves, subscribes to the, "objectivist," Ayn Rand philosophical school of thought. https://salemcenter.org/research/objectivism/
The market has dipped considerably not because there has been a change in state of the outcome - it has still been established based upon all of the source data from Chinese CDC reports, that there have been over 100,000 new covid cases per day average over a 7 day time period, as discussed at length below.
What people are betting on now, and why this market has dipped to currently 38%, is they are betting on the person who seems to be in charge of this contest, based upon his initial blog post, Richard Hanania (as well as whoever he may be working with or who has been assigned to resolve the bets). The bet is now essentially, "What will Richard Hanania choose?"
So that being said, for your consideration:
https://twitter.com/RichardHanania/status/1622357411867815937?cxt=HHwWgsDTkY7e4oMtAAAA
> New York Times: We examined Mandarin-language obituaries over 4 years of the Chinese Academy of Engineering & the Chinese Academy of Sciences to get an estimate of Chinese COVID deaths. Conservative media: Wow did you see that one guy who DIED SUDDENLY?
Further he goes on to say:
> The NYT is not a charity, it makes a profit because its audience wants to pay for real news. Conservatives don’t, they like conspiracy theory documentaries and political rants that tell them what they want to hear.
There's a funny comment on his Twitter thread, this did not come from me:
> Hanania isn’t arguing that NYT data is correct. What he is saying is that they are doing quantitative analysis on tangible figures to come up with estimates, as opposed to ritualistic hocus pocus nonsense that conservatative mouthbreathers are doing.
As I gain more of a nuanced understanding of what Salem is and who Richard Hanania is, at least from his Twitter profile...I don't necessarily see him automatically stating that the Chinese government did not report >100,000 cases any more for purely ideological reasons. I think he seems to be someone who actually may respect quantitative analysis and facts. If you look through his Twitter feed, he makes fun of conservatives a lot, as well as liberals.
So if you are on here betting purely because, "China lies, Salem Center is Conservative, they will Resolve NO purely because they recognize China as being a country that tends to lie in general," - I would perhaps reconsider your bet. According to what, "China Government reports," have put out, which goes in line with the market description, and doing an analysis as I have below, as well as looking at the WHO analysis which has been pointed to by the Johns Hopkins analysis...we are above the 100,000 range by multiple factors.
Why would Salem and why would Richard Hanania throw away his normal quantitative approach? Think very carefully about that - don't just assume the conservative position.
@PatrickDelaney with all due respect, I think you are *very* confused; charging others with the hypothetical idea that "Salem is conservative, so they tend to think that China lies in the *millions* of reported cases, so they will prefer to resolve this as ... less cases" is irrational mental gymnastics not grounded to any reality (real or imagined) whatsoever. People here are not idiots, regardless what they seem to believe about the eventual outcome of this market; notice that, after I posted a clear, succinct, and emotionless argumentation, the price fall has actually stopped. I (and you, I trust) would very much prefer this argument of mine to be the very *first* thing anyone checking comments here will see; so please, if you cannot contribute anything productive to the discussion, as it seems, kindly avoid *cluttering* it.
@IraklisTsatsoulis Indeed.
@PatrickDelaney What exactly do you think we have been betting on for the past couple of months? It has been obvious since the announcement of the end of 0 COVID that China was going to have 100,000+ cases, the question was if they were going to report them, with “report” defined in such a way that it showed up in OWID. This has always been wheat we were betting on, and the fact that you would like this to be a market on the true case numbers, rather than a particular measure of the reported numbers, has no bearing on the outcome of the market.
@IraklisTsatsoulis and @CharlesPaul - China has reported more than 100,000+ new cases per day on a 7-day average basis, end of story, check my comments below. I'm not saying anyone is an idiot, I am saying they are wrong. This market should have already resolved as yes.
@MatthewLilley Ha ha...as he goes and buys over S$ 800 of discounted shares at 23%. You're getting played. Anyway...I stand by what I said. Not confusing at all. 100k+ cases per day from government reports - still not a single person has event attempted to invalidate this and it's now just arguments based upon technicalities.
@SalemCenter kindly consider the following:
1) According to OWID, "Raw data on confirmed cases and deaths for all countries is sourced from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University." (https://ourworldindata.org/explorers/coronavirus-data-explorer , tab "Sources")
2) CSSE has just posted an announcement, clarifying that "the World Health Organization is receiving updated COVID-19 case totals from the People’s Republic of China National Health Commission (NHC)", and that "our team has concluded that we will not be integrating case or death totals reported by WHO into our data repository" for reasons of their own that have nothing to do with the purpose and essence of this market. CSSE concludes with "We recommend accessing data for WHO after or around December 9, 2022 when the difference between our time series begins expanding." - https://github.com/CSSEGISandData/COVID-19/issues/6576
3) Taken together, the above suggest that the resolution source has effectively become unavailable, and that the unavailability clause should be activated - a suitable alternative should be found.
4) The data reported in the WHO portal, coming from the PRC government, is such a suitable alternative: https://covid19.who.int/region/wpro/country/cn . Only thing that needs to be done is to subtract the Taiwan cases, since, according to the CSSE announcement referenced above, they are also included to the *official* numbers reported by PRC (inclusion of the Hong Kong and Macau SARs is not an issue for the purposes of this market, since they are parts of the PRC).
@IraklisTsatsoulis
1) OWID is not legally bound to continue sourcing from JHU, maybe they'll change their mind. (Since JHU CSSE Data repository is stopping *all* live reporting on March 10th, this is not totally implausible).
2) "for reasons of their own that have nothing to do with the purpose and essence of this market" is false; they're not integrating the WHO data because it's incomparable to the data they'd been collecting (based on the inclusion of Taiwan, Hong Kong and Macau, and other reasons stated at the link you provided), and of course part of the purpose and essence of this market is about not using incomparable data.
3) JHU CSSE is not unavailable, they're going to continue updating China COVID cases according to their past procedure (at least until March 10th).
4) The question here is not whether Hong Kong and Macau are part of the PRC; the question is whether the previous OWID dataset included Hong Kong and Macau. Based on JHU CSSE's comments here, sounds like the answer is no.
@SamDittmer
"maybe they'll change their mind" - yeah, but until they *will*, they are unavailable.
"JHU CSSE is not unavailable" - true, but JHU CSSE is *not* our source, OWID is. And judging from them (OWID) constantly reporting 0 values since Jan 16, they do not even bother to report the *non-zero* values reported by CSSE.
Hong Kong and Macau can be handled similarly to what I have suggested for Taiwan; it will not change the picture.
@SamDittmer
1) Irrelevant. OWID is no longer reporting for China (recording 0 because you no longer have a source available is not "the official data says 0"), and there is an alternative available that is reporting govt data, namely WHO. See "was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used." The govt reports exist, the numbers are public, OWID is not carrying them.
2) Irrelevant, and untrue as it pertains to the question. Anyone can subtract Hong Kong and Taiwan (and perhaps Macau too) off manually to yield China.
3) Irrelevant - China data from OWID is unavailable, that is what matters.
4) On its own merits, still insufficient. The numbers WHO reports look like they would still reach 100k/d even if 100% of Hong Kong and Macau were assumed to have been infected over the course of a week, and generally an individual is considered incapable of repeat infection within a certain duration (30 days etc, I suspect WHO has a standard it uses if China does not).
@SamDittmer In fact, OWID reporting constantly 0 since Jan 16 while CSSE *does* report non-zero values was already a smoking gun regarding unavailability.
@MatthewLilley Quoting from the JHU CSSE git issue: "There is no means to continue to update Hong Kong SAR, Macau SAR, and Taiwan using their independent reporting and include WHO data for China that would not result in individual cases or deaths being duplicated." So it's not true that it's obvious how to subtract off those numbers.
@SamDittmer The only not-obvious thing here is... why they do not proceed themselves with the subtraction, since they do have independent reports for these areas!
@SamDittmer I am aware that JHU CSSE say this, but what point are they actually making? China reports X, for Mainland China + Macau + Hong Kong + Taiwan. We have separate counts for Macau, Hong Kong and Taiwan. Ok, subtract them off! Is the issue that we don't know what numbers China is using for Macau, Hong Kong and Taiwan (in which case the question arises how this was ever resolved in the past - I presume it was that the numbers explicitly reported by geographical area), and any guess "might" be wrong? Because the obvious guess is they would use the same numbers those jurisdictions report. (For Hong Kong and Macau the Chinese govt presumably has full access to the data, for Taiwan it's all a bit made up... does the Chinese govt pretend it has "the real Taiwanese numbers" or does it just copy Taiwan?)
In any case, the scale of the numbers is so huge that even with the most outlandish assumptions it does not matter. WHO lists China as having 40 million cases in the week of Dec 19. You can subtract off the entire populations of Macau, Hong Kong and Taiwan and there is still a million cases a day left over! (This is of course not how one *should* approach this - unless there is a very good reason to do otherwise, just take the Chinese number and subtract the independent reports for these 3 other locations. But there is plenty of room for a lower bound that is still way above 100k/day).
@MatthewLilley If your metric here is official government reports, I think this report https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_13141/202301/t20230125_263519.html , which has been mentioned elsewhere, is pretty incontrovertible that the Chinese government is publicly stating that it had > 700k positive tests per day at one point in December, which gives you a > 100k/day seven-day average. I'm not contesting that.
And if Salem said "The intent of this market was always based on government reports, so we'll accept that report and resolve as YES", I wouldn't have any complaints - I've lost plenty of fake money on getting in late to the ups and downs of this market anyway, sunk cost "fallacy" is actually just true apparently.
But I still think the most plain reading of the market criteria is that for OWID to be "unavailable", they have to explicitly take their website down or say "We are no longer reporting live numbers", not just report a long string of zeros, and the Dec 17 update doesn't overrule that.
As to why JHU CSSE or OWID makes the decisions they do - I have no idea, except I suspect the fact that JHU CSSE is stopping all live COVID reporting on March 10th means that they think that tracking cases or tracking government reports of cases is no longer worth the effort, and so decided that making a fair decision about the old China numbers was also not worth the effort.
@SamDittmer
"for OWID to be "unavailable", they have to explicitly take their website down or say "We are no longer reporting live numbers"
I am afraid this is a *way* too limited definition of "being unavailable", and too many people here seem to agree with this definition.
@SamDittmer Thanks for that. I actually hadn't seen that document (I have $0 in this market currently and have been paying attention only intermittently). So that clarifies what we are disagreeing about.
Count me in as someone who thinks that is way too limited a definition of "unavailable" - I think inputting 0s each day even if your source is printing non-0s is "we are no longer reporting live numbers", just without saying it.
Under your definition of unavailable, there is absolutely no point of ever making the Dec 17 clarification. Given they did, I find your definition of unavailable tenuous.
But perhaps more importantly, I am also of the view that extreme litigiousness over interpretations rather than a sensible degree of discretion is a good way of killing prediction markets. The focus should be on making accurate predictions about the world, not being psychic about when a random website decides to stop printing the data. "Rules" provide a veneer that there is no arbitrariness / subjectivity in which way outcomes are judged, but outsourcing arbitrariness is potentially much worse... and rather dumb if everyone acknowledges that the benchmark (that the question is framed as interested in) has been reached.
@MatthewLilley The Dec 17 clarification is meant to establish a necessary, not a sufficient condition for market resolution, based on certain arguments made in this thread that sufficiently reliable case estimates alone should lead to the market resolving as yes.
I agree that extreme litigiousness is undesirable; this is why I think it's beneficial for markets to have resolution criteria described as precisely as possible, and for the resolution to follow those criteria based on a narrow interpretation. Another reasonable option is to do what a similar market on Manifold does, and say explicitly "This market will resolve based on X's best judgment", but then you risk having the market recreate the blind spots of person X. https://manifold.markets/ACXBot/14-will-there-be-more-than-25-milli
With the benefit of hindsight, I think the best argument for betting "Yes" in, say, November 2022 was "China's going to have a case explosion, and even if they lie, the case numbers will be so high their lies will not be credible and they'll eventually walk it back" and the best argument for No was "China has staked so much on zero COVID that it would look like a massive defeat for them to end zero COVID and admit a massive case explosion at the same time, so they'll find some way to 'declare victory and get out.'" (I can't speak to the Yes bettors, but the No argument was essentially my rationale back in November). And both arguments turned out to be correct. So I agree that making or losing money in this market does not tell us much at all about how reliable someone's internal model of the world is in general.
@IraklisTsatsoulis "In fact, OWID reporting constantly 0 since Jan 16 while CSSE *does* report non-zero values was already a smoking gun regarding unavailability."
This is a misunderstanding - OWID data for "China" has always been for mainland China (all provinces excluding Hong Kong and Macau), whereas the JHU CSSE dashboard for "China" shows data for all provinces including Hong Kong and Macau.
The CSSE data for cases by province is here: https://github.com/CSSEGISandData/COVID-19/blob/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv
It does in fact show zero daily new cases for mainland China since January 10, and this is what OWID reports.
It is non-zero for Hong Kong (until January 29) and Macau, and OWID also reports this: https://ourworldindata.org/explorers/coronavirus-data-explorer?zoomToSelection=true&time=2022-09-01..latest&facet=none&pickerSort=asc&pickerMetric=location&Metric=Confirmed+cases&Interval=New+per+day&Relative+to+Population=false&Color+by+test+positivity=false&country=CHN~MAC~HKG
Johns Hopkins is ceasing efforts on integrating WHO data into their own repo. This means that Salem's clause goes into effect:
> Should that source be unavailable, a suitable alternative will be found.
Johns Hopkins recommends the following in their recent post:
> Cases and deaths data reported by WHO can be found here: https://covid19.who.int/region/wpro/country/cn. We recommend accessing data for WHO after or around December 9, 2022 when the difference between our time series begins expanding.
So again, if you look at the week of December 19th, 2022 WHO has, "new cases" at 29,376,927, which - divided by 7 is 4,196,703 ... over 4M new cases per day reported from the China NHC via the WHO.
Aside from this, Johns Hopkins states their forward operational posture:
> Update deaths based on totals provided in the China CDC novel coronavirus infection briefs (https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/)
Which, as I had mentioned previously, from this report:
https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_13141/202301/t20230115_263381.html
Which shows 1.27 million hospitalized cases of new coronavirus infection for that week, which, back-dating those cases into a 7-day average linearly you get:
* 1/6/23 +211,666
* 1/7/23 +211,666
* 1/8/23 +211,666
* 1/9/23 +211,666
* 1/10/23 +211,666
* 1/11/23 +211,666
* 1/12/23 +211,666
I don't see anything from the above justifying a sell-off other than the expectation that Salem is going to actively choose to improperly resolve this market. What appears to be happening is people betting on Salem actually changing their criteria from, "was reported by the government," to, "we don't like how the government reported these numbers."
This goes back to the original point which is that Salem's entire contest is predicated upon the concept of, "non-experts being able to predict things better than experts using proper incentives." So the incentive they have set up appears to be, "what outcome makes Salem happy?"
Basically, anyone betting NO is betting on what Salem's intellectual bent is, rather than the reality of the situation, which contradicts the purpose of the contest which is allegedly to, "be objective." https://www.cspicenter.com/p/introducing-the-salemcspi-forecasting
Again, I plead with anyone betting, "NO" to falsify the idea that the Chinese government did not indeed report 100,000 new covid-19 cases on a 7-day average, without talking about what Salem wishes to happen - what am I missing?
I should correct my above by saying: Basically, anyone betting NO is betting on what they believe Salem's intellectual bent is (rather than what Salem's actual intellectual bent is). Though I would confess I agree with the NO betters that this is indeed Salem's intellectual bent...ha ha.
@PatrickDelaney Thanks for the update. The data you've shared over the previous weeks has been very helpful in solidifying my NO position, so I'm grateful for that.
@PatrickDelaney Instead of irrelevant pleads for irrelevant issues, let's try to be productive on the one issue that matters here: we now have an official announcement from JHU CSSE (it would be arguably useful to provide the relevant link: https://github.com/CSSEGISandData/COVID-19/issues/6576) that 1) China does provide *official* data to WHO, which are reported by the WHO portal, and 2) that they will not include them into their own portal for technical reasons. I have seen people here claiming that OWID get their data from JHU CSSE, but is this somewhere formally *documented*?
@SpencerHenderson I'm not hearing any disagreements with what I'm putting out there though in terms of the ground truth of the situation.
@PatrickDelaney If yes, then the latest announcement offers some solid ground to call for the unavailability clause to be activated, in accordance also to the later clarification about "government reports" and "official data". Everything else is unproductive grumble...
@IraklisTsatsoulis Yes, if you go to the OWiD github and the data GUI page itself, they list JHU CSSE as the source.
So you go here:
https://ourworldindata.org/explorers/coronavirus-data-explorer
Then navigate to, "sources" and it says:
Data published by COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University
Link: https://github.com/CSSEGISandData/COVID-19
Then you go to that repo and read through the issues and announcements.
@IraklisTsatsoulis Based upon your comment I'm realizing that I should probably provide references and links for everything I say. I am assuming that people are following along with this market and the comments I have made and research I have posted over the last couple of months. Obviously having been obsessed with this issue, I have a lot more background information. I'm hoping that Salem at least has a tally or a procedure for going through things.
Hopefully my above comments clears things up a bit in terms of the original source for OWiD. If not, happy to provide more background.
@PatrickDelaney Based upon standard practices, you (and all of us) are supposed to provide references when quoting what *others* say.
@IraklisTsatsoulis Sure, absolutely. What I mean is - I should be doing that on every post. Basically, I need to provide a huge blog post on every post it seems. Are you able to see a trail of my past comments? There are sources to what I am claiming in there, going back through December. But that being said, I realize that I am making the claims so the onus may be on me to compile what I have written.
(re-compile what I have written, that is)
I would like anyone from the NO camp to explain how they think there were less than 100,000 new COVID cases per day on a 7-day-average basis from Chinese government reports? I would love to be falsified and taught something new. I am begging anyone to take a sincere shot at falsifying what I have put forward here without talking about OWiD or using the word, "proxy," or anything like that. I would like to read something really interesting.
If they were gunna resolve YES based on {XYZ reasons elaborated on by others}, they would’ve already. Idk Salem’s thought process here, but it’s v much cleaner to go “the link we provided doesn’t show 100k on X date.” Vs “ok so we did a meta composite analysis of various sources of various degrees of pedigree and consensus shows….”
I could be wrong, but I’ll take 4:1 on this bet if y’all keep it in the 80s.
@NathanKrummen Put in a limit order then.
@ConnorPitts already got 1k share :]
I ain’t about to say this is a lock in my favor, but at this price I feel good taking the risk.
@NathanKrummen Why stop at 1k then?
So there seem to be two main schools of thought on this bet:
1. Absolutists, who believe the bet we're all betting on should have 100% to do with whatever shows upon the OWiD graph. They contend that it would not be fair to use any other source and ignore the December 17th, 2022 update paragraph.
2. Instrumentalists, who understand that the OWiD graph is merely a feed and believe that the market is already as good as resolved, as evidenced by the market riding at 80%, the WHO, actual CDC China reports, etc.
The absolutist argument amounts to, "Oh please give me a handout Salem, I really need to be proven correct because my feelings will be hurt if not." Not very much in the spirit of the free market!
Another funny part of this is...people who advocate for this market resolving as, "NO" and want to ignore the Dec 17th update are basically advocating on behalf of the CPP who has maliciously severely undercounted COVID cases at this point, which - I am sure being in the Salem Center community they love that thought. "Oh, no, there wasn't over 100k cases! Oh, no, we need to go by the original China case counts when they stopped reporting cases because it was so embarrassing, those are the gold standard!"
@PatrickDelaney I think you’re strawmaning the absolutist argument here. The issue here is that prediction markets try to measure underlying reality (COVID cases in China) by means of a proxy (CCP reported data aggregated by a third party.) it was obvious from the beginning that the proxy variable was not a perfect match for the underlying reality, someone asked Salem about this, and they answered, as is their right, that if the two divulged, the proxy variable would be used to resolve the market. I agree this is somewhat of a weird choice, given the potential availability of better proxies, but it is their right to make. Why can’t they just change it? The problem here is that a prediction market resolution criterion is essentially a contract between Salem, the people betting no, and the people betting yes. Yes, it is not ideal if the clear language of the contract behaves in a way not intended by the signers, but the reason why the courts won’t let you out of a contract just because it had unexpected results is that, if this were to happen, contracts would no longer be reliable. The key issue here is that if Salem were to change the resolution criterion now, then no one could trust them to remain consistent in the future, and markets would go from honest attempts to predict (sometimes imperfect) proxy variables to attempts to guess whether Salem was going to change the rules in the middle of a market. If you don’t like it, you are more than welcome to create your own markets and short circuit the entire problem by making the resolution criterion “Will I (Patrick) have a mean estimate > 100,000 for the highest 7 day rolling average of Chinese CVOID cases before 3/1.” This avoids much of the proxy variable problem, at the cost of being less predictable to people who don’t know how you think, thus reducing the volume of people willing to trade on your market. Salem, for valid reasons, did not go this route, and to change course midway through would destroy trust both in them specifically, and in the system as a whole. I apologize for the length, but this debate has gone on long enough that someone needed to formally state the absolutist position, and I didn’t have the time to make this more parsimonious.
@CharlesPaul Salem's betting contest is adjudicated based upon the principle that, "Regular folks can be better than experts at predicting outcomes given market incentives." It's not a court of law. What would destroy trust in Salem as an institution would be if they arbitrarily not abide by that free market principle. So in this case, the proxy, OWiD is no more than an API, pointing to Johns Hopkins. This has always been clear and visible on the OWiD website. Johns Hopkins, the team that maintains the data source for China cases, which OWiD feeds from, has stated that the WHO data should override the Johns Hopkins data in the interim. Keep in mind...I didn't say that...they did! I didn't pick the Johns Hopkins feed...Salem did! Salem's own judgement criteria has resolved in the YES direction. This is an instance where their own market, scoring at 80% is reflective not only of the ground truth, but also what their proxy is reporting.
So to use your court of law analogy, what Absolutists would have us believe is that if you went into a court of law with a contract that has a pretty picture at the top, but then there are actual words at the bottom of the contract that say, "this is based upon source A and we should always defer to source B," the judge should look at the pretty picture and disregard source A. That's not how contract law works in any jurisdiction on earth.
@PatrickDelaney What is actually happening here is that Salem is saying “We will go by whatever OWID says” OWID says “We are based on JH data” and JH says “out data is unreliable, you should use WHO data.” The question of where OWID gets its numbers is important, but ultimately it is those numbers we are betting on. You are probably right that OWID will update eventually (I agree with that, I’m just here as a speculator taking what I acknowledge as a high payout low probability bet) If the contract says “resolution is based on this number over here) and the people who maintain this number over here say things about their number, those are interesting, but do not change the fact that this number is still the contractual resolution criterion.
Also, I know this is not a court of law, but the same principles still apply, because the best practices for contract resolution are going to be similar whether you are a judge or a private arbiter (as Salem is here)
Finally, you say “ This is an instance where their own market, scoring at 80% is reflective not only of the ground truth, but also what their proxy is reporting.” Salem obviously cannot use the current probability of the market in determining whether to resolve it, because a. 80 ≠ 100, and if they were to resolve this one because it is at 80%, then they would have to resolve every other market at above 80 or below 20 for the same reason b. Because then someone could bid the market up or down enough, and then tell Salem “look at the market, you need to resolve now, it’s basically a done deal” and c. Because that is circular, and would cause the market to bet on how likely it was to dip above a point, which would take it further from its true probability.
@CharlesPaul First off, thank you for laying it all out there and bringing up new points. That being said, respectfully, Salem already said in an update, "This market was never meant to estimate the true number of cases in China, but was always to be based on government reports." then they further say, "any alternative source to Our World in Data will have to rely on official numbers to be used." If you read over the super long thread which has been going on for months now. 1. I first established that case loads were really high and stated that China is going to end up reporting something within the realm of reality because it's just ridiculous. People did not listen to me and continued to bet NO. 2. I next established after objections from many, that government reports did in fact report numbers, and that it was likely they were going to report higher numbers due to international pressure. Folks again did not listen to me and continued to bet NO. 3. I established that, while one might argue that the government reports might not be discernible because they are in Chinese, may be using case definitions that get lost in translation, ultimately we're going to have to interpret case numbers through a window of expertise, and that expertise has already been selected as Johns Hopkins. People continued to not listen and bet NO. 4. Now that everything I have said so far has come true, people continue to not listen and win on a technicality, which is that - the API reporting the source data is the ground source of truth, which it has been made abundantly clear by Salem that it has not. I'll give another update: 5. Johns Hopkins is trying to figure out how to fit their previous longitudinal figures in between three different scenarios: whether to treat new hospitalized cases as equivalent to new scenarios, whether to set that as a secondary source, e.g. set up a new column for going forward, or whether to just use WHO data. They are also thinking about how to avoid double counting. Either way, it has been established that cases have gone up above 100k 7-day average and it has been reported by the government (China CDC), no one debates that, it's a matter of data transformation. So I would say, if your betting strategy is to be contrarian because of a larger possible payout, that's respectable, but it's also because you know it's not representative of what's going on, it's betting on a fluke.
@PatrickDelaney that’s a whole lotta gymnastics for an 80% bet.
@NathanKrummen how so?
@CharlesPaul Let me kindly point out that @SalemCenter has never said "“We will go by whatever OWID says”. To keep the contract analogy, from the very beginning there was a clause "Should that source be unavailable, a suitable alternative will be found.", plus an amendment clarifying that the market "was always to be based on government reports". There is a strong argument here that, by reporting simply *zero* values (this is an important detail - OWID does not merely underestimate, they have *stopped* reporting cases whatsoever), the source has practically become unavailable, and since government reports have unambiguously become available through the WHO, the clause should be activated.
@IraklisTsatsoulis reporting 0 cases is still reporting a number, and the details are clear that if China stops reporting this resolves as no.
@CharlesPaul China, specifically the Chinese CDC did not stop reporting. China already reported >100,000 on a 7-day average so it should resolve as YES.
@PatrickDelaney The key here is that you made a good argument that OWID will eventually show 100+ cases, but your prescience so far doesn’t mean that we are “winning” by having the market remain unresolved. I could go make a great argument in the “will China invade Taiwan” market that China will invade Taiwan on March 15th, but no matter how good that argument is Salem should not resolve the market based on that argument. The fact that the market is not yet resolved is genuinely reflective of the fact that no resolution criterion have been met. Remember, the key issue we are currently betting on is whether OWID will ever be updated to reflect the real case count. As for “betting on a fluke” I prefer to think of it as betting on the propensity of the CCP to keep on officially lying about case counts, even if everyone knows its a lie. I look forward to seeing which one of us turns out to be correct.
@PatrickDelaney OWID begs to differ
@CharlesPaul No, you're wrong that this is a prescience argument. The China CDC has already reported over 100k cases per day, that's a fact. They came out with a big report on January 25th, 2023 and this is evidently where WHO aggregated its data.
https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_13141/202301/t20230125_263519.html
I mean they are reporting millions of positive tests cases per day...millions, much less 100k.
As I have said previously, the world would be a much easier place to understand of the a-holes were always wrong, or always lying or whatever, but unfortunately that's not the case. It's very possible of course that these reports still grossly undercount or fudge what was actually tested. It's also possible that China is just very disorganized as it still has much less development and infrastructure than you would think. But anyway, what we do know is that the report above speaks for itself...which is again why the WHO COVID dashboard shows millions of case counts per week.
Repeating Salem's words above:
> Thus, any alternative source to Our World in Data will have to rely on official numbers to be used.
@PatrickDelaney we’ll see. I personally think the market is close to fairly priced now, but if you disagree you are more than welcome to buy it up. The nice thing about prediction markets is that comment section arguments are objectively resolved, even if the resolution uses metrics which are not always ideal.
@CharlesPaul Yes, you are correct that Salem may resolve this improperly. I would buy more but I'm already all-in. Care to loan me any?
@CharlesPaul Just so we are clear, reporting 0 is NOT reporting a number. That is just a programmatic default now that OWID is not receiving any data.
The map is not the territory. This contest was supposed to be about finding the best predictor of reality, not the best predictor of what would happen with OWID because OWID was just an attempt to provide resolution criteria. If the site has stopped updating, it should clearly be ignored for other sources of data. Damningly, no one here who has argued that this should resolve no has even suggested that the cases are under 100k. You are all just trying argue about technicalities.
@MatthewB Precisely.
Futher update... https://github.com/CSSEGISandData/COVID-19/issues/6520#issuecomment-1411292936
Directly from Johns Hopkins:
> We're working to bring this data into our repository but need to rectify some conflicts with data for Hong Kong SAR. For now, I recommend observing the WHO dashboard for mainland China COVID-19 data.
It's over. The source that Salem has been using has recommended using the WHO data. The WHO data points to an average of ~5M per day during the week of December 19th, 2022. @SalemCenter please resolve.
WHO is now reporting a daily average of 40M new confirmed cases during the week of December 19th, 2022 in China, that's a 7-day average of ~5M per day for that week. https://covid19.who.int/region/wpro/country/cn ... Salem should resolve as yes based purely upon this and all of the other information I have put together.
@PatrickDelaney I don't understand the case for why this should resolve positively now. The resolution criterion isn't respected.
@HenriLemoine There has been an average 5M new COVID cases per day using a 7-day-average measurement from China CDC reports. The OWiD points to Johns Hopkins, which has said to point to the WHO.
@PatrickDelaney I know that. However,
> This market resolves as YES if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023. Source for settling the market will be Our World in Data at the link below.
> https://ourworldindata.org/explorers/coronavirus-data-explorer
> Should that source be unavailable, a suitable alternative will be found.
Since that source is not unavailable, we can use it. If you open that url and look at what the resolution criterion is asking for, it is not respected. I don't get why any of what you're saying is relevant to the current resolution. I find it a valid argument for why it will eventually resolve positively when the website is updated. Still, there's no way it should resolve positively now, is there? The resolution criterion doesn't say, "this will resolve according to what whoever OWiD trust says", and even less, "this will resolve according to who the source used to inform the OWiD points to". What am I missing?
Essentially, I see this question as reducing to how soon the website will update with better data. But the update def hasn't happened yet!
@HenriLemoine Well just wait. Suppose there is no update in the data until March 1 (but it's obvious an update is still being hashed out) and then they backdate "X million cases a day in Dec/Jan". That should satisfy the criteria right? The relevant date is date of reported cases, not the date the update occurs (and by corrolary, if they announced "actually 5 million people were infected in Feb 2020", surely this would/should not qualify).
@PatrickDelaney This question relies on OWiD which is showing zero cases per day as of Feb 2. Unless OWiD as a source is no longer available, then this is still resolving NO as of now per a direct reading of the rules.
@HenriLemoine Why are you cherry picking only one part of their entire description and not including the December 17th, 2022 update?
@SpencerHenderson Can you please give a real reason rather than just stating arbitrary rules that you wish would happen?
@MatthewLilley Ultimately it's up to Salem. It would be incredibly hilarious if they resolved this market as that distanced from reality after they put out all of this information and blog posts about how non-experts are better than experts at public health...but then again I guess that would truly solidify their reputation as a complete intellectual prostitute and make their entire contest meaningless.
@PatrickDelaney "Source for settling the market will be Our World in Data at the link below."
I'm relying on the actual rules, you're substituting a different source not mentioned in the rules.
@SpencerHenderson What is OWiD?
@PatrickDelaney Our World in Data. The source used to resolve this question
@PatrickDelaney You’ve been making variations of this argument for months now. If Salem found it convincing they would have already resolved the market. So presumably they don’t agree with you. I don’t know that continuing to harp on it will change that.
@JosiahNeeley Salem knows that if they resolve the market, "YES" then that could put me pretty high up on their leaderboard in their little contest. At the same time they know that I'm on to their little political game, they don't want me up there...it's all very funny. We'll see what happens. I guess we're all really going to see what they think about the free market aren't we?
@PatrickDelaney I have a fairly strong prior that people with persecution complexes are wrong
@SpencerHenderson OK, and what is Our World in Data exactly?
@AAh That's a really good prior!
@PatrickDelaney Google and/or basic reading comprehension are your friends here
@SpencerHenderson Hah, nice use of insults. OK let's be adults. Please explain to me what OWiD is in detail. Do you not want to discuss it?
Welp, I guess I'm out as far as this competition goes. It was nice knowing you all. Even if I sold I'd be so far behind that I can't win anyway.
@RobertGrosse I tried to get you to switch buddy, sorry.
Ironically, this comment ended up chosen as "Proven Correct". Can't say I've ever been more glad to be proven incorrect!
Salem's Data Source now shows 59,895 deaths having been reported by China on January 15th, 2023.
https://ourworldindata.org/explorers/coronavirus-data-explorer?facet=none&Metric=Confirmed+deaths&Interval=New+per+day&Relative+to+Population=false&Color+by+test+positivity=false&country=~CHN
However, case totals have not been updated yet.
This is because Johns Hopkins received death reports from the China CDC that was in the equivalent format to how they had in the past. However, case reports are in a different format, quote:
We are still working to understand how we may incorporate the "hospitalized cases of new coronavirus infection" into our dataset. We appreciate your patience.
https://m.weibo.cn/status/4860354171766047
China CDC Chief Epidemiologist has stated on Weibo that 80% of population has already been infected with COVID. https://weekly.chinacdc.cn/en/article/doi/10.46234/ccdcw2020.175
New Report from China CDC Today retroactively pointing back to 1/15/2023. Now it's a matter of time before this percolates into Johns Hopkins, which will then percolate forward into OWID:
https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_13141/202301/t20230115_263381.html
So basically here the original source of the original feed is reporting 1.27 million hospitalized cases of COVID, with 104,018 severe cases.
> 1. Cases
> On January 12 , 2023 , 31 provinces (autonomous regions, municipalities directly under the Central Government) and Xinjiang Production and Construction Corps had 1.27 million hospitalized cases of new coronavirus infection , and 104,018 severe cases (including critical cases) , of which 7357 were severe cases of new coronavirus infection 96,661 cases of severe basic diseases combined with new coronavirus infection . From December 8 , 2022 to January 12 , 2023, medical institutions in 31 provinces (autonomous regions and municipalities) and Xinjiang Production and Construction Corps accumulatively had 59,938 hospitalized deaths related to COVID - 19 infection. 5,503 cases died of exhaustion, and 54,435 cases died of exacerbation due to underlying diseases combined with new coronavirus infection .
@PatrickDelaney The problem is that the 1.27 million is the number of people in the hospital in total, rather than the number of *new* cases, so it isn't clear how that can or should be integrated into the figures.
@RobertGrosse I think this should actually update us in the direction of NO. Like they announced previously, this is the first report without any numbers on cases, so it’s only hospitalizations now. OWID also updated to now include just a bunch of zeros for cases in China:
https://ourworldindata.org/explorers/coronavirus-data-explorer?time=2022-11-10..latest&facet=none&Metric=Confirmed+cases&Interval=New+per+day&Relative+to+Population=false&Color+by+test+positivity=false&country=~CHN
IANAL, but seems like that plausibly could be enough for a resolution as per the update on Dec 17 @SalemCenter
@1941159478 and @RobertGrosse ... If you assume 7 days with 0 previously reported, then it's 1.27M/7 = 181,428 new cases/day reported over a 7 day average.
You guys have been super insistent that we use OWID with no context, no wiggle room, again and again. Now if the logic doesn't favor you, we have to all of the sudden go back on that.
Just stop and think to yourself these very simple principles:
* We know there are huge numbers of cases in China in the millions, and even the large new figures are under-reporting.
* We know that China has already reported millions of cases on a provincial and municipal level.
* We know that COVID expands exponentially, that China has a huge population, and that Chinese New Year is on Sunday, January 22nd, it's a huge holiday with tons of opportunity for spread
* We now know there are updates on the CDC level in the 7 figure magnitude.
* We also know there is massive international pressure for China to report more within a range of reality.
Rather than continuing to dance around the topic, sell your NO position while you can. Probability is not in the NO direction.
@SalemCenter let me remind you that the wording on your update was, "...as long as there was never 100,000 cases reported for any 7-day rolling average. " and "...any point between the opening of the market and February 28, 2023. " The word ANY is used twice and signifies ANY 7-day rolling average within the time period of the market. So mathematically, we have to wait until at least 7 days after market closing to be able to resolve as NO. Whereas to be able to resolve as YES, it only has to go above the threshold at one point.
@PatrickDelaney Why would you assume that everyone in the hospital was infected in the last week? If you assume they were all infected over the last two weeks, you then get 1.27M/14 = 90,714 cases per day.
Based on e.g. NYT hospitalization data vs cases, the former *might* be more plausible than the latter, but it certainly isn't conclusive. It's another way to indirectly estimate the number of cases per day rather than China directly reporting it.
This doesn't meet the terms of the Salem Center update because the hospitalization data doesn't correspond to time-stamped infection data, and so it gives us no information about 7-day rolling averages.
@PatrickDelaney I still think that this will probably resolve based on OWID. I just don’t think the China CDC’s report you linked will be incorporated there. Like @SamDittmer , I don’t see how they could take that report’s numbers to get a number for daily confirmed cases.
Also, I made another market on the main Manifold based on this question that will resolve based purely on the OWID confirmed cases numbers, no wiggle room. So we can bet there without worrying about ambiguous resolution criteria: https://manifold.markets/1941159478/will-our-world-in-data-show-100000
@1941159478 Again, I'm trying to tell you guys, I'm not just being an a-hole here trying to win at all costs, I'm not that kind of guy, never did well in sports, etc. All right, so let's say you want to go with OWID. OWID is an automated feed...it feeds off of Johns Hopkins. Johns Hopkins has a discussion thread / issue on Github about this. They may place the numbers at 1.27M new cases on January 12th, depending upon how they interpret it. Then that will percolate forward into OWID, which is completely an automated feed off of Johns Hopkins. Johns Hopkins may use a 14-day window, that puts us above 90k for the 31 Provinces. Combine that with Hong Kong at around 20k last time I checked, we're over 100k. Now let's say Johns Hopkins doesn't aggregate the Hong Kong cases in there for some reason, as the WHO does them separately, then we're at 90k, but it's going to be easy, with Chinese New Year, to add additional cases.
The only reason I would be thinking about selling my shares would be if the model in my mind for China was that it was an absolute North Korea, does not respond to International Pressure, etc., or if I specifically had it out for China, rather than trying to be an impartial observer of some kind who, "doesn't know." Honestly there are so many reasons to believe China is still super under-counting, but...that doesn't mean that's the most accurate way to model how China behaves in our minds. I don't want to be a demagogue, all of these international affairs can be very opaque and misunderstood.
@PatrickDelaney I wouldn't count on Lunar New Year putting you over the top, even in that scenario. Keep in mind that the 1.27m figure is *down* from the peak in December.
@RobertGrosse ... how do you know that? I thought there weren't any official reports from China? ;-) ... Yeah, you could be right. This article from Nature estimates the same, based upon the report from the Henan Province Health Official that I had linked earlier. https://www.nature.com/articles/d41586-023-00075-4 ... The official from the CDC who has been contributing on the Johns Hopkins Github China issue stated that these sources can be used in non-longitudinal analysis, which Nature seems to be extrapolating from as well. They also mention Baidu search. I'm not sure how to do a Baidu trends analysis, but the China trend search for COVID had a peak second to last week in December. That being said, if you look at past Google trends for the USA compared to when actual peaks were happening, they seem to go off, as it's an informational measure rather than a true measure. https://trends.google.com/trends/explore?date=all&geo=CN&q=%2Fg%2F11j2cc_qll,%2Fm%2F01cpyy
OK, more news. I think this market is over now, officially reported as a YES.
This market is done. Over 2.86 million patients reported on December 23rd, 2022 by Jiao Yahuai the NHC.
Take 2.86M/100,000 = 28 days, which would be November 25th as a start. We know the ramp-up started after November 25th because we know COVID Zero Ended around December 8th, 2022, so the new cases per day average assuming uniform distribution would have been around 143,000. This means the peak 7-day average of new cases would have been far greater, in the hundreds of thousands.
https://www.reuters.com/world/china/air-travel-recovers-china-amid-covid-infection-worries-2023-01-14/
> China reports huge rise in COVID-related deaths after data criticism
> Between Dec. 8 and Jan. 12, the number of COVID-related deaths in Chinese hospitals totalled 59,938, Jiao Yahui, head of the Bureau of Medical Administration under the National Health Commission (NHC), told a media briefing.
> Of those fatalities, 5,503 were caused by respiratory failure due to COVID and the remainder resulted from a combination of COVID and other diseases, she said.
https://www.cnn.com/2023/01/14/china/china-covid-deaths-intl/index.html
> Jiao, the medical official, said fever clinical visits and Covid hospitalizations in China have already peaked.
> According to the NHC, fever clinic visits – both in cities and rural areas – have been declining since the peak when more than 2.86 million people visited them on December 23, 2022.
> On January 12, 477,000 people visited fever clinics across China, Jiao said Saturday.
> The NHC said hospitalizations of Covid-19 patients also peaked on January 5, 2023, when 1.63 million people was hospitalized, and 1.27 million Covid patients were still in hospital as of January 12, Jiao added.
@PatrickDelaney Technically the 2.86m is people visiting clinics on an individual day (Dec 23) - that date being the peak. (Unclear from the CNN story if they published a time series or just those two days of data mentioned). Also technically, these aren't cases. Of course, 100k per day would have been cleared easily - it's really just about what the CCP allows to be printed.
@MatthewLilley It's also absurd to claim different numbers *for the time period that the official source already covers*. In order to get this to resolve YES, you first have to start with a really tortured interpretation of "Should that source be unavailable", since as long as the market uses the source specified, there is no way it can resolve YES.
@RobertGrosse Tbh I'm not very sympathetic to that pov when the "specified" source is printing utter tripe. Rules are not always better than discretion. But also, sources are back-revised all the time. I would expect the 10x increase in deaths to flow through to the OWID site at some point, for example.
@MatthewLilley They already used their discretion and stated very clearly in the Dec 17 update that they will follow the "utter tripe".
@RobertGrosse Actually, to be precise, they stated no such thing. What they stated was "...was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used." I won't pretend that is definitive either way, but it certainly doesn't support your statement.
What counts as official numbers/government reports? "Jiao Yahui, head of the Bureau of Medical Administration under the National Health Commission (NHC)" detailing specific numbers at a press conference sounds like it would meet that criteria. (And i presume at some point that these numbers will turn up in a pdf or on a webpage somewhere).
Doesn't mean they will necessarily use this. But it is hard to argue it is ruled out by the update given on Dec 17.
@MatthewLilley Is OWID still available? If so, nothing else matters, per the resolution criteria. The "alternative sources will have to use official government data" is only describing what would happen in the event that OWID goes down for some reason.
@MatthewLilley "Visits to a fever clinic" are not the same as "confirmed COVID case", and number of hospitalizations does not come with a time stamp saying when those corresponding cases were confirmed. If you open the door to allowing substitute metrics based on other statistics mentioned in a press briefing, that's a very wide door.
I've made these arguments before below https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#91FQeyokKmnvul1oXsE8, but there's a pretty convincing case to be made that total "true" US cases have already passed the resolution criteria for that market as well. Where's your indignation at the "utter tripe" from the NYT?
@SamDittmer You seem to think this last part is a killer argument. It really is not. Is there any other remotely major country that essentially stopped testing/counting/announcing case counts right during the middle of a major wave? In the US (or insert other country of your choice here) the fraction of infections that are reported as cases has been highly stable in the short/medium term. It's apples and oranges.
""Visits to a fever clinic" are not the same as "confirmed COVID case"". Lol, I said this in my first comment above.
Hospitalisation: technically you are correct (for now), empirically it's clutching at straws. You don't get 1.63m people in hospital at one point in time without 700k cases over a week. (Average hospitalisation duration is not going to be that long). In fact, if Covid hospitalizations at time t and t+7 are x and y respectively, then y-x essentially provides an extremely conservative lower bound on cases (at least, if they deem you a case when they admit you to hospital as a Covid patient).
@MatthewLilley Re: Hospitalizations, if there were an official government release of a dataset including "confirmed Covid hospitalizations at time t", and there was a gap of 700k over a 7 day time period, that would be satisfactory evidence, to me at least, in favor of resolving yes.
Re: My "killer argument" and your "apples and oranges", I do not think a discussion of the "fraction of infections that are reported as cases" is a very clear answer. The ratio of hospitalizations to cases in the US has increased by a factor of 2 since last July. Salem Center created that market on August 5th, so should we double all case numbers from the US to account for the approx 2x slowdown in testing since August 1st?
I of course agree that China's undercounting rate shifted quite a bit more dramatically recently than the US's, but I think they are shifting in the same direction for basically the same set of reasons - popular exhaustion combined with institutional incentives against measuring and reporting on your failure to achieve your stated goals.
@RobertGrosse OWID is an auto-feed, literally nothing else, it means nothing. It feeds off of Johns Hopkins. Johns Hopkins tracks longitudinal data...they are not sure what to do because data changing has been reporting.
On the Johns Hopkins GitHub issue discussion for China data reporting, a guy from the US CDC chimed in and noted that for disease tracking you have longitudinal data and event data for outbreaks. They are both what they are.
If you have an event datapoint from an outbreak, say 10,000 ebola cases in a remote region, this is equivalent to saying there must have been past an X-day average of cases, just based upon pure easy 2+2 math. If you don't think that 2+2 is 4, then I can't help you.
*they are not sure what to do because data reporting has been changing. -- have not had my coffee yet.
The point is, you have to realize what a betting market is. It's not supposed to be, "ways to keep yourself stupid," it's supposed to be, "ways to make yourself more informed about reality by arguing with others." You keep bring ing up OWID. Why? This is not a pickup hockey game. This is not arguing about whether someone got cross checked or not incessantly. This is about continual information gathering. Please tell me something new rather than repeating that same tired line.
China's CDC has recategorized COVID as a category B disease, from what I can tell from Johns Hopkins discussions on Github. Following the discussion thread, an official from the US CDC chimed in:
> Hey, just a note from a lurker from US CDC. I think this was already mentioned, but the cessation of daily COVID-19 reporting by China CDC was planned based on the previous pronouncement of the downgrading of COVID-19 to a "Category B" disease. 1/8 was the final daily report.
> There's probably a worthwhile conversation to be had about how to proceed with most of the publicly available data sources drying up. I'll mention that, even in the absence of updated data from Mainland China, the ongoing collection of disaggregated data on HK, Macau, and Taiwan is of great utility for us and others, as these data are submitted in the aggregate to WHO via WPRO office.
I have asked the individual from the CDC what his thoughts were on disaggregated data from Chinese municipal and provincial level reports. Depending upon his response, I may or may not concede that my previous claims about 88.5M+ cases may not be valid, depending on the response. They are the expert, not me.
That being said, I still think the market needs to wait to close until after the official close date, because something could change between now and then in terms of national level reporting.
https://www.bbc.com/news/world-asia-china-64208127
> Provincial official Kan Quancheng revealed the figure - amounting to about 88.5 million people - at a press conference.
> Mr Kan did not specify a timeline for when all the infections happened - but as China's previous zero-Covid policy kept cases to a minimum, it's likely the vast majority of Henan's infections occurred in the past few weeks.
Let's assume linear growth starting from the end of COVID Zero on December 5th, 2022, so that's 35 days.
88.5M/35 = 2,528,571/day or 2.5M/day in one province accounted for by one Provincial-Level official in one province, Henan.
Extend that out to 90 days, well prior to the end of COVID-Zero, also assume linear extrapolation:
88.5M/90 = 983,333 or 983k.
How far back would we have to go for this one province to be at 100k per day using linear extrapolation?
88.5M cases /100k cases/day = 885 days
855 days / 365 days = 2.3 years, approximately Sunday, September 6th, 2020.
You would have to go back to 2020 and assume non-exponential growth of the pandemic to make the absolutely ridiculous contention that no government official has stated more than 100k per day. This market is grasping at straws to prove itself on its priors at this point which is baffling to me.
@PatrickDelaney
1. Obviously millions (if not tens of millions) of people have / are being infected with Covid daily in China.
2. I am sympathetic to the argument that a source like this *should* be sufficient for resolving "if China reports a 7-day rolling average of at least 100,000 covid-19 cases at any point between the opening of the market and February 28, 2023." (Since one province cannot exceed the entire country).
3. Indeed, I bet heavily long on the >20k cases per day market (and made a lot of profit), and bet heavily long on this market too. I thought the Chinese govt would be unable to stop the spread (and contrary to common narrative, I think they only gave up on lockdowns because they realised cases were exploding anyway, without meaningful hope of containment).
4. In terms of "prediction markets having value", predicting the true state of the pandemic in China seems far more important than overly-literal predicting "what data will the central Chinese govt print as then funneled through website X". However the people who run the mkt seem to have decided on the latter.
5. Since the people who run the mkt weren't swayed by previous reports from a provincial govt official of millions of cases a week (which immediately would suffice), I don't see why this would be newly swayed by this either. In fact, if they were, it seems like a massive change of mind over what counts and what doesn't. Trying to predict that seems pointless to me.
@MatthewLilley Re: (4), a prediction market on the "true state of the pandemic in China" only has value if there's actually some credible way to determine the true value, which would require the government of China to impose a more aggressive testing regime than anything that occurred during its Zero Covid days (and you'd still get error bars there due to false positive and false negative rates of the tests). See also my comment below: https://salemcenter.manifold.markets/SalemCenter/china-reaches-100000-covid-cases-by#q5oJq8KcvHG5s3RsffhO.
There are no comments on the US COVID market https://salemcenter.manifold.markets/SalemCenter/covid-peak-of-over-300000-by-end-of making similar arguments, even though of course the "true" number of COVID infections in the US is also much higher than the reported numbers, due in part to a set of political choices about the availability of and requirements for testing.
You may respond that the China numbers are *obviously* more manipulated than the US numbers, but part of the claimed value of prediction markets is to avoid the need for these types of judgments and guard against bias manifesting in selective demands for rigor.
Re: US case numbers, based on the NYT they peaked around 800k during the Omicron wave in January 2022, with a corresponding hospitalization rate of 8.2/100,000, while the current hospitalization rate is around 3.4/100,000. Assuming the hospitalization rate has remained constant, this gives us an estimate of 330,000 cases per day, and that's still ignoring the fact that the 800k cases per day in January 2022 was itself certainly an undercount due to lack of testing.
@MatthewLilley "Prediction Markets," is a misnomer. It's a betting market. We're betting on an outcome. The outcome has been determined according to the betting judge's criteria. The betting judge is going to either continue with its criteria, maintaining trust, or it's going to continually raise the bar because it wants a particular outcome, or it wants to pick a particular set of winners, it has friends betting a particular direction, etc. It's very clear what the Government has reported over 100,000 new cases, per day and that the YES bet should be arbitrated as a win, it's a question of how professional Salem wants to be.
OK, so if you go and look at Our World in Data now, you will see a massive spike up to 25k cases in December where it had previously shown close to zero cases.
https://ourworldindata.org/explorers/coronavirus-data-explorer?facet=none&Metric=Confirmed+cases&Interval=7-day+rolling+average&Relative+to+Population=true&Color+by+test+positivity=false&country=~CHN
Looking at my discussions below, I have been insistent that the cases being reported have been over 20k while people on the, "NO" side have been insisting that they have been alternatively in the low thousands, around 2k or around 0.
Of course you will see a massive drop-off, followed by another growing spike. This has to do with data reconciliation between how various sources report data, translate the data, and a trend in countries doing data reconciliation and data coming in late.
I still believe that, given the massive actual number of cases in China, this feed may go well over 100k at one point once the data is properly reconciled.
In addition to that, province-level and city-level health reporting, which still classify as Government reports, have already reported in the millions, so this market should have already resolved as YES based upon those government reports and the December 17th, 2022 clarification clause:
"...any alternative source to Our World in Data will have to rely on official numbers to be used."
Which opens the door to alternative sources from different officials. Since the national definition of COVID verified cases was changed on December 14th, 2022 to, "symptomatic cases only," we have to be reasonable and accept additional data sources beyond the monolithic previous metric set as being Our World In Data.
@PatrickDelaney if the OWID site is updating again, that SHUTS the door to using other sources.
@PPPP Why?
@PatrickDelaney Plan A for the data is still operational, so why would we go to Plan B?
@PPPP > This market was never meant to estimate the true number of cases in China, but was always to be based on government reports.
So after having looked at Johns Hopkins discussion on their github repo and some reports on what's going on with China's reporting ... 1. The Chinese CDC is looking to come out with an update on January 8th, 2023, so we'll see where that lands. 2. I want to make it clear that there *already has been well over 100,000 daily cases reported from China.
https://www.nbd.com.cn/articles/2022-12-24/2607048.html
From December 24th, 2022:
> According to Qingdao Daily, on the 23rd, Bo Tao, director of the Municipal Health and Health Commission, introduced that the current peak of the new crown infection in Qingdao has not yet arrived, and is in the stage of rapid transmission before the peak. According to the monitoring data, the current daily number of new infections in Qingdao is 490,000 to 530,000 people, and it will increase at a rate of 10% tomorrow and the day after tomorrow .
So that's an official government figure -- the director of Municipal Health of Qingdao, a very large city in China, stating that the current daily new infections is between 490,000 and 530,000.
That right there should resolve the market as YES under these two criteria:
> Should that source be unavailable, a suitable alternative will be found.
> This market was never meant to estimate the true number of cases in China, but was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used.
These above are official numbers, from one City. There are official numbers in the hundreds of thousands from other cities. Johns Hopkins is trying to figure out how to contextualize this, but basically...the market should have already resolved as YES.
More information about NBD - National Business Daily, this is a State-Owned Newspaper. https://en.wikipedia.org/wiki/National_Business_Daily
All right so from discussions with a NO better, @Sam Dittmer , I have been told the following, re: Salem Center:
> Re: Salem Center coming up with an "actual expert", I don't think any special expertise is required to understand that "Our World in Data or a suitable alternative" means "Our World in Data or some other dataset that counts the same thing."
OK, so that's fair, Salem Center may not have access to an expert. That being said, this opens the door to...an official source, originally pointed to by Our World in Data, such as Johns Hopkins, interpreting and giving context for China's reports...not just purely our non-expert opinion on the Chinese NHC, which, since we're not health experts, could contain some kind of mistranslation.
Let's look at Salem's above clarification on Dec 17th:
> ...If at some point China stops reporting data completely ... this market will resolve as NO
So according to Johns Hopkins, China has not stopped reporting data completely, they have changed the definition of a new case to a, "non-asymptomatic new case." In other words, if you are asymptomatic, you are no longer counted as of Dec 14th, 2022:
https://coronavirus.jhu.edu/region-data-notes
So now we are left with, "Should that source be unavailable, a suitable alternative will be found." ... so again, the fact that the Our World In Data API is no longer hooked up to the Johns Hopkins API, or that the Johns Hopkins API is not feeding out what they believe to be the actual case numbers at this time, because the developers have not made updates yet, has nothing to do with the final resolution of the question, which is, according to a NO better's own interpretation, again:
> "Our World in Data or a suitable alternative" means "Our World in Data or some other dataset that counts the same thing."
So let's look at John's Hopkins current count page. I can't put images in here, so you have to navigate through this page to look at the weekly total:
https://coronavirus.jhu.edu/region/china
Navigating to, "past week," we see: 164,182...divide that by 7, we are at 23,454 weekly average cases from China reports as of the date of authoring this comment. In summary, we are using:
1. Expert interpretation of the number of cases being reported by China,
2. With context, from Johns Hopkins, which was the original source for Our World In Data
3. Using the newly updated definition from China as of December 14th, 2022.
I would still say, based upon COVID's exponential growth, and China's population being 1 billion, and Sun, Jan 22, 2023 even with counting in the above way, we're going to easily hit +100,000 reported by China.
@PatrickDelaney OWID isn't "unavailable" though. As long as the OWID link works, then according to the resolution criteria, they should *not* choose an alternative source.
@RobertGrosse JHU and OWID are both reporting the same numbers once you subtract Hong Kong from the JHU numbers, i.e. they're both reporting the number zero.
@RobertGrosse I realize that you're voting NO and that's obviously what you want to happen, ha ha...but that's...not at all what the Dec 17th update says. Sam...no, we already went through this. You're being inconsistent. Below you're saying it's over 2k, now you're saying it's zero. That's inconsistent, you're just throwing things against the wall to see if it sticks now. It's 20k in Hong Kong, 20k+ in China...separate reporting, we already discussed this. Please refrain from bad faith argumentation.
@PatrickDelaney I don't think you can reasonably say the source is "unavailable" when it is still up. If anything, *you're* the one who is brigading and making bad faith arguments. It's clear that you really want this to resolve YES and are desperately trying to persuade Salem to do so, presumably for the sake of points on *a completely different site*, even though the resolution criteria don't leave much wiggle room to resolve it as anything but NO, should things continue as they are.
@RobertGrosse Hey Robert, honestly that's a bunch of, "others," - I became a member of Salem's Manifold Market under the terms and conditions of this website, just like you and it has no bearing on how this market should resolve. If you feel that threatened by my presence here...I wonder what others reading this thread might think about how confident you are in your research? If Salem wants to resolve this arbitrarily NO, that is absolutely their prerogative. However if they want to do some actual research and stick to the plan they have laid out, make it fair for everyone betting and not just make things up, I think it would be a lot better for Salem in terms of showing they are consistent. Yeah, I really do care about points on that other website, absolutely. Is there anything wrong with that? Why would I spend any time on any of this at all if I didn't care about points? Are you not also playing for points? It's a betting game, the objective of which is to determine a set criteria upon which a decision will be made. Brigading? Where is the rest of my brigade? Anyway... so continuing the ACTUAL conversation without resorting to ad-hominem attacks, let's see if I can persuade you to sell your NO shares and buy YES. ...So first off, 1. We're going off of Johns Hopkins Data, that has been established, and here it is: https://coronavirus.jhu.edu/region/china. 2. Case count is tracking *new* case count, as that is what the World-O-Meter site previously linked showed. 3. China switched from asymptomatic testing included to asymptomatic testing not included on December 14th per Johns Hopkins. 4. 7-day average Case count is running at 23,573 today. 5. We have 56 days left for the case count, including the significantly under-reported numbers to increase. 6. Historically, COVID has increased exponentially. 7. It's thought that the actual cases could be in the millions: https://www.science.org/content/article/china-flying-blind-pandemic-rages 8. Other countries are imposing testing on Chinese travelers now. 9. Chinese CDC reports over 35k according to that article, which is higher than the Chinese NIH, which Johns Hopkins uses, so there could be a reconciliation of 10k+ reports more. 9. WHO is exerting pressure on China to up its reporting. 10. Chinese New Year is on January 22nd, 2022 this year, over a month prior to this market ending.
Basically, if you're taking an objective look at things, a 26% resolution is super cheap. Unless you know something I don't know about Salem - I have stated in other analysis I think Salem is ultimately going to follow the criteria it sets out.
@PatrickDelaney If you're actually serious about participating in the contest and not just manipulating the resolution for your personal benefit on a different site, why not put your money where your mouth is and bet YES? If you're actually right about this resolving yes, that would be enough to catapult you into the top 10.
@PatrickDelaney As I explained in some detail below, the JHU numbers include Hong Kong, so no, that’s not 20k + 20k, you’re double counting the Hong Kong numbers.
The average for China (as defined by OWID) was 2976 from December 16th to December 23rd, per the China NHC numbers. It is zero since around December 25th, since they’ve reported no new cases since then.
This csv file is from John’s Hopkins:
https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv
@SamDittmer Point of clarification on a Github issue, referencing the Repo you just referenced: https://github.com/CSSEGISandData/COVID-19/issues/6414
> We are working to identify what 'relevant epidemic information' will continue to be released. At present, we have not identified any official or media sources that are publishing cumulative cases or deaths at the national or provincial level. Until a reasonable alternative is found, our data will remain stale.
@RobertGrosse OK done
@PatrickDelaney Interesting. The issue you linked links to https://www.chinacdc.cn/jkzt/crb/zl/szkb_11803/jszl_11809/202301/t20230103_263169.html, where it appears that they have started reporting cases again (4833 on January 2nd, going by Google Translate).
Of course, there's still the question of whether that will 20x before things are over or not.
@RobertGrosse To be fair from what I read and is linked above from Johns Hopkins, the China CDC never stopped reporting cases, the China NIH did and noted that reporting will transfer to the China CDC. However, it is not clear to me how often the China CDC reports. We know for sure that cases have more than 20X'd that figure, and other figures in the 20,000 range, in fact we have 2,000 X'd the 20,000 range by upper estimates. But yes, this bet as has been clarified many times, is on reporting.
This is why I would still advocate for a safer bet being on the, "YES" side...at some point, even though China is going to underreport, it's going to need to report something, and then that's going to filter through to a more accurate number in the John's Hopkins reporting. Johns Hopkins is not in the business of obscuring health data...they may go on the conservative estimate side after getting more reports from the reports they have been following in China, but not grossly conservative. Statements coming out in the news right now saying that, "China has stopped reporting," is not complete, and information claiming that they, "won't do so," is uninformed opinion, as there has already been accounts from many in the Chinese government stating that the number of infected is likely in the millions (see the Manifold.Markets discussion on this from last month).
Again, making another top comment here since there's a long thread below mine just arguing back and fourth providing no evidence, just noise.
I'll make this more explicit. The above data set linked, when you follow and read about it, is actually just a wrapper for the data set at:
https://github.com/CSSEGISandData/COVID-19
Data published byCOVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University
Which if you look at their commit on December 19th, 2022 here:
https://github.com/CSSEGISandData/COVID-19/commit/36796b264241680f10af5e8da25d899da6e35bea
This Github repo inexplicably deletes the National Health Commission of the People's Republic of China and the China CDC.
> - National Health Commission of the People’s Republic of China (NHC): http://www.nhc.gov.cn/xcs/yqtb/list_gzbd.shtml - China CDC (CCDC): http://weekly.chinacdc.cn/news/TrackingtheEpidemic.htm - qq.com: https://news.qq.com/zt2020/page/feiyan.htm#/ - National Health Commission of the People’s Republic of China (NHC): http://www.nhc.gov.cn/xcs/yqtb/list_gzbd.shtml - China CDC (CCDC): http://weekly.chinacdc.cn/news/TrackingtheEpidemic.htm
It's reasonable to accept that they have deleted those sources because the reporting from China is now insane.
So, that's the source that Salem points to as the end source, right? Well, China hasn't stopped reporting. Johns Hopkins has stopped reporting data from China . This is very critical to realize. Here you have an institution in the United States pointing at an Institution in China, and then later deleting their source data stream. So being that we are betting on what was pointed at in a previous data stream, namely the China NHC and the China CDC...we really only have those pieces of information to go off of, because that's what Salem put out there originally, with their question, "Will China report X" -- not "Will Johns Hopkins report that China reports X" Right? Follow me here?
So next let's look at what the old data stream from the China NHC is most recently reporting:
So you go here to this website...again, linked by Salem Center, not by me:
http://www.nhc.gov.cn/xcs/yqtb/list_gzbd.shtml
Then click on the most recent report as of the time of authoring this:
http://www.nhc.gov.cn/xcs/yqtb/202212/cb666dbd11864171b6586887c964791c.shtml
Toward the bottom of that document, translated, they report:
As of 24:00 on December 23, 31 provinces (autonomous regions and municipalities directly under the Central Government) and the Xinjiang Production and Construction Corps reported that there are currently 41,699 confirmed cases (including 515 severe cases), 350,117 cured and discharged cases, 5,241 deaths, and confirmed cases reported. There are 97,195 cases, and there are 24 suspected cases. A total of 15,377952 close contacts were traced, and 147,724 close contacts were still under medical observation.
A total of 9,161,081 confirmed cases have been reported in Hong Kong, Macao and Taiwan. Among them, 495,614 were discharged from the Hong Kong Special Administrative Region (108,371 discharged and 11,373 deaths), 1851 in the Macao Special Administrative Region (1521 discharges and 19 deaths), and 8,663,616 cases in Taiwan (137,742 discharges and 15,039 deaths) in Taiwan.
So that provides the number of cases on that specific day, from midnight to midnight, with the number including Hong Kong + Mainland (31 provinces) being 108,371 + 97,195
So to calculate a 7-day average, you have to look at the previous 7 days and average up the totals from all 7 days. The number has increased from the 23rd from the days prior.
Looking at the 16th, as a starting point, you have:
http://www.nhc.gov.cn/xcs/yqtb/202212/6a7d9897b43e42b3903bc067f8154789.shtml
31 provinces:
76,361 cases
Hong Kong:
105,143 cases
Now again compare to Dec 23rd:
http://www.nhc.gov.cn/xcs/yqtb/202212/cb666dbd11864171b6586887c964791c.shtml
31 provinces:
97,195 cases
Hong Kong:
108,371 cases
So, you can review all of the days in between yourself, they go up from 105,143 + 76,361 to 97,195 + 108,371 which is clearly well over 100k as a seven day average.
So again, THIS WAS THE DATA SOURCE LINKED IN THE ORIGINAL MARKET.
THE WAGER SAID NOTHING ABOUT JOHNS HOPKINS, IT WAS ABOUT *CHINA REPORTS*
If this were a legal contract, if the Salem Center cares about Rule of Law at all, the data reporting would be based upon the feed from the original link they provided, particularly given that they even clarify things on December 17th, and I quote:
> This market was never meant to estimate the true number of cases in China, but was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used.
The best official numbers are the ones that Our World in Data was originally using, but simply cut off based upon the API being cut off by Johns Hopkins.
I really don't think it's worth my time to spend any more time on this...you put together a prediction market because you want better access to research and what's going on in the world. You get someone, me, who has spent a lot of time in China, research this for you. Now it's up to you what you want to do with the facts or whether you want to just resolve it however suits your agenda. Good luck with your, "research institution."
@PatrickDelaney You're conflating "confirmed total cases", and "new confirmed cases". We go from a total of 376,361 cases (not 76,361) on December 16th to a total of 397,195 cases (not 97,195) on December 23rd, for a total of 20,834 cases, or approximately 2,976 cases per day over that 7 day period.
As other people have noted elsewhere, the original linked page at Our World in Data uses the language of "new confirmed cases" to describe what it means by "7-day rolling average", so it is new confirmed cases we want, not the cumulative totals.
Also, FWIW, the Our World in Data page distinguishes between China (i.e. the 31 provinces), Hong Kong, and Taiwan for the purposes of its reporting, so it seems natural to assume that the intended metric was the 31 provinces, not the 31 provinces + Hong Kong + Taiwan. But adding Hong Kong + Taiwan to the above calculation does not get us close to 100k per day.
On a separate note, I appreciate your efforts tracking down the various Chinese government reports from the CSSE repo, and I find your animosity towards Salem Center rather excessive for a disagreement over the correct mechanism to parse governmental reports.
@SamDittmer OK, so it sounds like the interpretation of daily cases is based upon some standard figure that I disagree with, that's fine, I understand that's how it goes, and I have to go by popular interpretation, I'm not going to dispute that.
Let's look at the WHO report, what they are showing:
https://covid19.who.int/region/wpro/country/cn
So that's around 26k new daily cases for now, which matches up with one of the figures within the reports I'm linking to. You are now currently saying that it's 20k, "total" cases out of the blue, arbitrarily. How do you explain the discrepancy between the official government reports, linked to by Salem Center originally, showing 26k, and lining up with the WHO reports, and indicating, "7 day average of new cases." Wouldn't that show that there are in fact 26k cases, not 2976, which is basically wishful thinking? Additionally, if you just Google, "Hong Kong covid cases," they display Johns Hopkins data, and show around 20k daily cases. So totaling those together, we're at 40k new cases per day being reported by China.
Unless Salem can come up with an actual expert on this, it's really just winging things...which is frankly surprising for an institution that bills itself as a policy center. How are you going to write policies if you don't know what's going on? I wouldn't say this is animosity, this is just like...the basics. I'm a random schmo doing far more research on this topic in my free time than Salem, an institution.
There are other concerns on this market, because of Salem's unclear point:
> This market was never meant to estimate the true number of cases in China, but was always to be based on government reports. Thus, any alternative source to Our World in Data will have to rely on official numbers to be used. If at some point China stops reporting data completely, this market will resolve as NO as long as there was never 100,000 cases reported for any 7-day rolling average.
So basically, the update demonstrates how Salem is going to resolve NO, but does not provide any scenario by which it's going to resolve YES. This just tells me they are waiting to resolve NO, and don't want to do any further research, they just want to accommodate to their pre-determined conclusion on the topic.
@PatrickDelaney I appreciate your willingness to yield to the "popular interpretation" here, but I don't think there's any ambiguity. I went to Our World in Data and saw how they described the 7-day rolling average. This market has always been based on metrics reported by Our World In Data, and any suitable alternative should be *measuring the same thing*.
Likewise, the Our World in Data dataset does not count the Hong Kong or Taiwan datasets under the China total (see e.g. August 10 2022 at OWID where the Taiwan and Hong Kong numbers are higher than the China numbers), so any alternative that Salem Center selects should also not include those numbers.
The reason the JHU and WHO datasets have higher case numbers than the OWID dataset is that these datasets do include Hong Kong. For JHU CSSE, go here: https://www.arcgis.com/apps/dashboards/bda7594740fd40299423467b48e9ecf6 and click on China and then Admin 1 and you can see that Hong Kong's numbers are included in the total.
So when you get 40k cases by adding together 20k and 20k, you're counting the Hong Kong numbers twice, when they should be counted zero times. The 2976 number is not "wishful thinking", it's division: (397,195 - 376,361)/7 = 2976, and that's using the numbers from the official government sources that you linked to.
Alternately for JHU CSSE you can download this csv: https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv which has cumulative cases by region up through 12/31/22, and you can confirm that in every region besides Hong Kong, the cumulative total has not changed since 12/20/22 or so.
Re: Salem Center coming up with an "actual expert", I don't think any special expertise is required to understand that "Our World in Data or a suitable alternative" means "Our World in Data or some other dataset that counts the same thing."
I am confident that the market will resolve as YES if the Chinese NHC (or any other official Chinese government sources tracking confirmed cases) shows an increase in 700,000 confirmed cases for the non-Hong Kong non-Taiwan China numbers over any 7-day period, whether or not this is tracked on OWID.
I assume someone at Salem is keeping half an eye on the official numbers coming out of China and the discourse in these comments, and feels that their Dec. 17th update was already sufficiently clear to answer any concerns. I don't know what sort of "further research" you want them to do. There's no new data coming out of the 31 provinces.
OK, so we have established that, "Our World in Data or some other dataset that counts the same thing." is the gold standard, got it. So this basically opens the door to...an official source, originally pointed to by Our World in Data, such as Johns Hopkins, interpreting and giving context for China's reports...not just purely our non-expert opinion on the Chinese NHC, which, since we're not health experts, could contain some kind of mistranslation.
@PatrickDelaney Yes, I think it would be completely reasonable for Salem Center to announce they'll use the JHU CSSE data and sum over the 31 regions of China counted by the OWID dataset.
I do not think it would be reasonable to suddenly start including Hong Kong numbers in the total count, since OWID doesn't.
I have done a great deal of research into this topic. This needs to resolve YES, otherwise Salem Center or whomever is in charge of this account is basically untrustworthy in terms of resolving markets. Please see my comments on this on this derivative market on Manifold. I am no longer participating in this market because I am unsure ultimately if you are going to resolve properly, but I think you should take a look. https://manifold.markets/AndyMartin/will-china-reach-100000-daily-covid#cxUnWbNP5I5RVD7TOah5
@PatrickDelaney It would be incredibly unfair if they resolve it contrary to what they explicitly promised in the Dec 17 update.
@RobertGrosse it is incredibly unfair to those of us who bet before the 17th if it isn't resolved positively. This is a prediction market. The prediction was whether China would report at least 100k cases as a weekly average. Chinese officials are acknowledging that the cases are far beyond that level. What is the point of this contest if the market doesn't resolve positively?
@MatthewB That is not what the Q says. A quote in an article from an official is not the same as a national daily report.
@PPPP Point out the substantive difference from an epistemic perspective. The report was for the purposes of having an authoritative source. I could even accept the argument that politics should be taken into account, so if they were falsifying the report, the market still resolves no.
Instead, it is universally acknowledged that the number is far above the target. Give me a good reason that a contest, which was supposed to find the best predictor, should reward the people who bet no instead of yes
@MatthewB “ Source for settling the market will be Our World in Data at the link below.” and that site is a compendium of the official govt reports.
“ I could even accept the argument that politics should be taken into account, so if they were falsifying the report, the market still resolves no. ”
I don’t see much difference between falsifying the data and opportunistically stopping the reporting. Both are govt decisions to hide the true number of cases.
@PPPP They effectively did both, since official case counts plummeted before they stopped reporting entirely, presumably due to greatly reduced testing.
@PPPP They didn't opportunistically stop. They stopped because the numbers were clearly inaccurate and useless once they stopped compulsory testing.
I'm still waiting for a good reason.
@MatthewB you got one. It’s been clear from day 1 that this market is based on a specific data source and isn’t just Salem picking whatever methodology they think best.
@PPPP No, I'm still waiting on you telling me the epistemic difference. And who is a better predictor between someone who bet yes and someone who very no. Don't tell me that you gave me one when you pointedly ignored both points.
@MatthewB one is predicting the official Chinese gov’t report totals, one is predicting the true # of cases. There’s the epistemic difference.
The better predictor is unresolved because the market is unresolved.
@PPPP How is that an epistemic difference?
@MatthewB There is a clear difference in epistemic status between the number of cases confirmed by a laboratory test and reported by a government agency and an estimate of the number of people who newly acquired (or re-acquired) a certain concentration of the SARS-CoV-2 virus in their body on any given day. The first number can be known exactly, the second number cannot.
This is why the Wikipedia page for the 2009 H1N1 swine flu pandemic (https://en.wikipedia.org/wiki/2009_swine_flu_pandemic) distinguishes between:
- Confirmed cases: 491,382 (lab-confirmed);
- Suspected cases: 700 million to 1.4 billion (estimate).
The OWID data source refers to "confirmed new cases", and so refers to the first sort of number, the number of lab-confirmed cases. OWID says explicitly "Due to limited testing, the number of confirmed cases is lower than the true number of infections." A Chinese official at whatever level of authoritativeness saying "we are confident the true number is at least X" is a statement about suspected cases, not lab-confirmed cases.
The number of confirmed cases is always a lower bound on the number of true cases (unless the false positive rate and testing rate are both high enough), and so any statistical estimate of the number of true cases will always be higher than the number of confirmed cases.
Some really weird trading patterns on this Q...
@AlvarodeMenard I have no idea what Dylan Levi King is doing. My guess is that they decided they have no chance of winning and just wanted to screw around manipulating the prices to see what would happen. But whatever is going on, at least it's free money for the rest of us.
For reference: This market got mirrored on the main Manifold including some spirited discussion on the resolution criteria.
https://manifold.markets/AndyMartin/will-china-reach-100000-daily-covid#2NFIlKtOTPKbMq58Ofx5
A note on the National Health Commission website says: "从即日起,不再发布每日疫情信息,将由中国疾控中心发布相关疫情信息,供参考和研究使用。" "Effective immediately, there will no longer be reports of daily epidemic numbers, and the Chinese Center for Disease Control and Prevention will report relevant information for reference and research purposes" http://www.nhc.gov.cn/xcs/yqtb/202212/7272431ee60c4f4d953b0c16257c230e.shtml. The CDC will still report items like serious cases, deaths, etc., I assume (it doesn't make clear what the "relevant information" might be), but not daily new infections, since testing has been mostly abandoned.
This is not an end to all data reporting, but an end to reporting data relevant to this question.
There's reporting on this in English, too, I see: https://www.asahi.com/ajw/articles/14801703. No more daily infection and death totals.
@SalemCenter: Does this fit "China stops reporting data completely"? https://jp.reuters.com/article/health-coronavirus-china-cases/chinas-national-health-commission-to-stop-publishing-daily-covid-figures-idUSB8N32X004
There are no more numbers. "China stopped releasing daily Covid-19 caseloads from Sunday..." https://www.scmp.com/news/china/science/article/3204531/china-stops-declaring-daily-covid-cases-wave-strains-hospitals-funeral-services
Nearly 37 million people in China may have been infected with Covid-19 on a single day this week, according to estimates from the government’s top health authority
https://www.bloomberg.com/news/articles/2022-12-23/china-estimates-covid-surge-is-infecting-37-million-people-a-day
Seems like this should resolve yes.
> Half a million people a day are being infected with Covid-19 in a single Chinese city, a senior health official has said, in a rare and quickly censored acknowledgment that the country’s wave of infections is not being reflected in official statistics.
>
> A news outlet operated by the ruling Communist party in Qingdao reported the municipal health chief as saying that the eastern city was seeing “between 490,000 and 530,000” new Covid cases a day.
> ...
> The report was shared by several other news outlets but appeared to have been edited by Saturday morning to remove the case figures.
seems like https://www.theguardian.com/world/2022/dec/24/chinese-city-seeing-half-a-million-covid-cases-a-day-local-health-chief could qualify as "official numbers" re: 12/17 update ?
@AndyMartin Agreed. And if the question is COVID cases, and some Chinese provinces report "all" of them, but the central govt either suppresses the data and/or only includes a subset, the sum of provincial reports would still seem to qualify per the question.
“Almost 250 million people in China may have caught Covid-19 in the first 20 days of December (12.5 million per day), according to an internal estimate from the nation’s top health officials…”
https://www.cnn.com/2022/12/23/china/china-covid-infections-250-million-intl-hnk/index.html
@AlexanderTippett Look at the update
Looking at the official numbers for the past seven days (I'll put them below), the average sits around 2200. This is local transmissions, not including cases imported from overseas, which have averaged 58 a day over the same period. So, the numbers given on https://www.worldometers.info/coronavirus/country/china/ seem to match this, and it conveniently rolls the local and international together.
Wednesday Dec. 14: 1,944 https://english.news.cn/20221215/6e560fe55c2f4966beb595c4d2aa37ce/c.html
Thursday Dec. 15: 2,091 https://english.news.cn/20221216/bf4fa1f5e7be433e80539a6ddecaad1a/c.html Friday Dec. 16: 2,229 https://english.news.cn/20221217/bcc317c90fef4745b8d5c25395890b95/c.html
Saturday Dec. 17: 2,028 https://english.news.cn/20221218/ed749defcf244c5d861997ebbdbf6473/c.html
Sunday Dec. 18: 1,918 https://english.news.cn/20221219/1139bd00f3d042daab3986b4c6a1df56/c.html
Monday Dec. 19: 2,656 https://english.news.cn/20221220/04774984bb86452b8043f1636f55b9db/c.html
Tuesday Dec. 20: 3,049 https://english.news.cn/20221221/0fe176f6416643f39523e876e364b943/c.html
https://www.worldometers.info/coronavirus/country/china/ remains accurate (it lags by a couple days), checked against official numbers on the National Health Commission of the PRC website, http://www.nhc.gov.cn/xcs/yqtb/list_gzbd.shtml. Although new cases are creeping up slightly, single day totals haven't broken 4300 and the rolling seven-day average is below three thousand.
These official new case numbers are fictional, so it doesn't matter whether or not actual case numbers peak. They are unlikely to go up between now and the Spring Festival travel season, which will begin in middle January. They will likely remain low during that period, since it would be a great inconvenience to start cranking official case numbers during the period many are traveling.
@SalemCenteer: I think this may be doomed to resolve on a technicality in one direction or another, but given the wording of "Should that source be unavailable, a suitable alternative will be found" I think there's an argument that can be made that this should resolve positively if (a) the OWID chart continues to not post updates through the rest of the surge and (b) there is widespread agreement China has reached 100k daily cases.
For example, testing is not mentioned in the description, and while it does use the phrase "if China reports", it doesn't make any distinction between symptomatic & asymptomatic.
Could we get clarity on how this will be handled?
Before we get a ruling, I want to weigh in. The key language here seems to be "if China reports..." (not "if it is widely acknowledged..." or "when we can assume..."). Whether or not OWID updates, the official reports are still coming from China. We know the total today was 2,229, for example: https://english.news.cn/20221217/bcc317c90fef4745b8d5c25395890b95/c.html.
This page has an updated 7-day rolling average for daily new cases: https://www.worldometers.info/coronavirus/country/china/. It's at 2346.
@DylanLeviKing yeah that’s a totally fair point.
The counterpoint I’d make to that is 100k is so small relative to total population that if China reports the peak is over in February and estimates more than 10% of the total population has been infected, this is equivalent to saying cases crossed 100k. It seems like this should meet the “if china reports” bar.
(90 days at 99,999 cases per day is only ~9M total cases)
@AndyMartin An update has been posted.
@SalemCenter thanks!
The link we're using seems to not have updated since the 13th, but the 7-day rolling average, taking up to the Friday number (2901), would be around 6600. There is no longer an official count of asymptomatic cases. The number of tests being taken has dropped to nothing as they are no longer required for everyday life, and everyone is sick of taking them.
I'm fairly certain China will reach 100k cases. I'm very skeptical about whether the Chinese government will report 100k cases though.
@AlvarodeMenard yep, they've already been cutting back testing.
Here's what the Chinese government is saying: "A speech by government adviser and former National Health Commission official Feng Zijian in which he predicted that the first wave could see about 60 percent of the population infected..."
https://www.washingtonpost.com/world/2022/12/09/china-covid-restrictions-relaxed-infections/
Just vibing here, seems a good thing to gamble on
@AlexRadcliffe lol
https://www.ft.com/content/cc8502e1-a59b-4df2-a392-2fec4ab72edd
> Guangzhou has partially lifted a weeks-long lockdown, a departure from the strict enforcement of China’s strict zero-Covid policy despite the city of 18mn people suffering its worst Covid-19 outbreak since the pandemic erupted.
> Officials in the southern manufacturing hub on Wednesday eased restrictions on movement over about half of the city’s 11 districts, including Haizhu, where migrant workers have clashed with police over the past month.
> Local officials’ announcement to ease lockdowns followed direct approval from Beijing, two people familiar with the decision said. The timing of the measures, coming despite nearly 7,000 new cases being reported on Wednesday, was seen as an indication of a broader change in policy direction.
Today's update brings this to 39,164 - with the recent protests/pushback (and continued lack of comparable growth in deaths), I don't think they'll be able to lock down tightly enough to consistently bring r0 below 0 country-wide for at least a few weeks, which is more than enough time for this to hit 100k in the current wave
Comparing the "relative to population" graphs of China against places like Singapore, Hong Kong, and South Korea shows there's so much room for China's absolute case number to grow: https://ourworldindata.org/explorers/coronavirus-data-explorer?time=2022-09-28..latest&facet=none&Metric=Confirmed+cases&Interval=7-day+rolling+average&Relative+to+Population=true&Color+by+test+positivity=false&country=CHN~KOR~HKG~SGP
I'm maxed out, but if I had more S$, I'd bet this up to 80-90%
@AndyMartin It seems to have rattled the planners enough to start massive lockdowns again. There will be no easing. I remain moderately confident this levels out below a hundred but potentially above fifty.
@DylanLeviKing yeah I agree with your point, although I think there's enough time between now and Feb 28 to where there will be two meaningful chances for a positive resolution. If deaths remain very low and only in very elderly people throughout this wave, it seems possible they will continue to ease and be slower to tighten when the next one starts.
@AndyMartin The situation could also be minimal easing and continued lockdowns but mismanagement, new strains, and mass testing sending the number up. I think the status quo will hold in enough other locations, but the PRD/YRD need to open to avoid further economic tanking. We'll see! I found this story from Shijiazhuang interesting: https://www.bloomberg.com/news/articles/2022-11-16/fear-and-panic-in-chinese-city-rumored-to-be-exiting-covid-zero
Play-money betting
Mana (S$) is the play-money used by our platform to keep track of your bets. It's completely free for you and your friends to get started!