Based on https://ourworldindata.org/covid-cases
This is question #14 in the Astral Codex Ten 2023 Prediction Contest. The contest rules and full list of questions are available here. Market will resolve according to Scott Alexander’s judgment, as given through future posts on Astral Codex Ten.
1,000
3.00
I just want to reiterate for awareness that the question is misstated, as per previous comments.
This question states: "Will there be more than 25 million confirmed COVID cases in China by the end of 2023?" That says "by the end of 2023" which means including previous years, unless you take a very unusual reading.
Scott's actual question is cases in 2023 only, not including previous years.
I strongly suspect many traders were misled by this mistake.
@ShadowyZephyr I do actually believe that the number of total cases in China in 2023 alone have already surpassed 25 million. But not the number of confirmed Covid cases in China in 2023 alone.
Scott posted a clarification here that says it is the number of cases in 2023, not the total number of cases in 2023 and earlier years.
You mean happening in 2023 right? Not if the total number will reach 25 million in 2023.
Yes.
Because the author of this question misunderstood and mistranscribed the question from Scott's original, I still think it should resolve N/A as per previous comments.
The original question text from Scott is
"14. Will there be more than 25 million confirmed COVID cases in China?"
which is understandably somewhat unclear, but based on the rules elsewhere the meaning of "in 2023" makes sense. The use of the word "by" was a mistake here on Manifold in adding the relevant context.
See https://manifold.markets/ACXBot/14-will-there-be-more-than-25-milli#WzHe1VojlXNmUTM7Mht6
@NuñoSempere My argument for N/A is that the Manifold question stated a completely different question than Scott's question. This is, in my view, a strong basis for N/A.
Also note that the link to the full rules is not working!
@jack Resolution methodology was specified in the comments question. From my perspective yes holders not reading this is their problem, for which they should rightfully suffer.
We can ask Scott Alexander to find the full list of rules if that is a problem.
I'm not sure I understand your argument. Scott clarified the resolution of Scott's question, that's not a problem anymore. I argue that the manifold question is still a problem, because the title "14. Will there be more than 25 million confirmed COVID cases in China by the end of 2023?" is fundamentally inconsistent with Scott's clarification.
I wrote out in my previous comments the full rules. https://manifold.markets/ACXBot/14-will-there-be-more-than-25-milli#WzHe1VojlXNmUTM7Mht6. Again, that's not the problem, the problem is that Manifold wrote a question that was unintentionally different.
@jack My argument is that title ambiguities is irrelevant, because the question details specify the method of resolution, which allowed for my high conviction trade back when this market opened.
@NuñoSempere In particular, if I recall correctly, Scott's full rules had a specific clarification that all questions should be understood to refer to 2023 only.
The rule is:
"All point questions are about the state of the world midnight on 1/1/2024, and all duration questions about the period 1/1/2023 - 1/1/2024, unless otherwise specified."
If you read the Manifold question (which is wrong!) in conjunction with that rule, I believe the best interpretation is that it's a point question (it's about whether an event happens "by the end of 2023"). That doesn't tell us whether cases before 2023 count or not, it remains unclear.
If you read Scott's question in conjunction with that rule, it's unclear, but I originally interpreted it as a duration question.
I pointed out the problem 5 months ago, in the comment I linked, and I argued that the Manifold question was probably wrong and it was probably supposed to count only cases in 2023.
The traders were wrong => They lose their money => This is priced in future questions.
The traders were wrong => They get a bailout => You create incentives for sensitive traders who cry for their mommy for an N/A when they are wrong. Or you get question creators create questions with built-in ambiguities that can either resolve as a surprising YES in their favour or as N/A.
I don't think it's valuable to the forecasting community overall whether traders are good at predicting one way or another in badly written questions where the question itself contained a misunderstanding.
The way I think about it, questions have an implicit context for how ambiguities are resolved, and that is priced in. On Kalshi, I would bet according to the legalese fine print regardless of whether it contradicted the title. On Metaculus, probably either the question will resolve as per the technical resolution criteria or it'll be resolved ambiguous, so either way you should predict according to the resolution criteria I think. On Polymarket, I expect it to be a coin flip whether the resolution will follow the spirit of the question or the resolution criteria, if they conflict, and I'd predict accordingly.
The traders were wrong => They get a bailout => You create incentives for sensitive traders who cry for their mommy for an N/A when they are wrong. Or you get question creators create questions with built-in ambiguities that can either resolve as a surprising YES in their favour or as N/A.
I 100% agree these are problems with badly written questions, but where I disagree is that I don't think it has anything much to do with N/A. If N/A weren't an option, then you could make precisely the same argument about resolving YES or NO. People in fact do argue a lot for their preferred reading of the question. And a malicious author can just as easily pick their favorite of YES or NO - that's much more common than N/A. Also, malicious authors can easily say they are going to follow their resolution criteria to the letter, and intentionally write a deceptive title, and that would be just as bad if not worse.
@BTE I saw that you matched my limit order. In case you want get out of this market completely, I just put in a much larger order at the same probability.
@1941159478 I have been thinking about this and it's odd that the Manifold question is phrased different from the Metaculus version. Technically the Manifold one can resolve YES while the Metaculus one cannot. You see what I am saying??
@BTE Exactly. They are phrased differently because the people who copied the question onto the platforms interpreted the ambiguous wording differently.
@jack I agree that it's notable, but I doubt they'll resolve differently; this market "will resolve according to Scott Alexander’s judgment, as given through future posts on Astral Codex Ten", so I expect it'll resolve based off of his intended criteria (i.e. on infections during 2023 alone).
The market is pretty clear that it's a mirror of his contest question, with its associated rules. It's just a shame that the rules were no longer accessible via the link (and that the questions were worded like this and transferred to Manifold without being changed).
(I'm saying this having been burnt after irrationally selling my ~10,000 NO shares at ~90% when OWID said they would incorporate WHO data on GitHub, due to a combination of silliness + forgetting the intended criteria)
@finnhambly I strongly believe this question should resolve n/a in case it was inaccurately transcribed from Scotts original question to here. Additionally, I think Scott should and is likely to judge it ambiguous