Poll: How likely are you to interpret a question "Will X happen?" to really mean "Will X happen by close date?"
Basic
12
Ṁ346resolved Jul 21
Resolved as
70%1D
1W
1M
ALL
If the question asks "Will X happen?" and doesn't make it clear what happens on the close date, how likely are you to interpret the question as "Will X happen by the close date?" (i.e. it resolves NO on the close date if it hasn't happened yet)?
As opposed to "Will X happen ever?" and the market does not resolve until either X happens or definitely cannot happen. If the close date is reached, either the market remains closed and unresolved, or the close date may be extended. In other words, the close date is irrelevant to the resolution, it's either a date to close trading in advance of possible resolution, or a guideline for when resolution is likely.
Vote by commenting with YES, NO, or a percentage at the start of your comment. (YES = always by close date = 100%, NO = close date is always irrelevant to resolution = 0%, you can also choose any percentage in between.) The market resolves to the average of valid votes in the poll as of 1 day after market close. You can update your vote by commenting again with a new vote; only your last vote will be counted.
A few examples:
https://manifold.markets/BTE/will-a-delaware-judge-order-elon-mu
https://manifold.markets/LarsDoucet/will-elon-musk-back-out-of-the-deal
https://manifold.markets/dreev/will-the-musk-twitter-deal-close (author later clarified their intent)
https://manifold.markets/Austin/will-manifold-set-abandoned-questio
https://manifold.markets/FRCassarino/will-i-quit-my-swe-job-at-metafb-to
https://manifold.markets/jack/if-i-take-paxlovid-will-i-have-a-co
This question is managed and resolved by Manifold.
Get
1,000
and3.00
Sort by:
Clearly, interpretations are all over the place and this is a big source of ambiguity. Authors can and should try to write their questions explicitly to avoid ambiguity, but ultimately the only way to truly address this is for Manifold to come up with some UX change that avoids this ambiguity. One possibility is replacing "close date" with an optional "resolution date" or something like that, but there are pros and cons to the various suggestions here.
@LivInTheLookingGlass You're always welcome to add your input of course, but it won't affect the resolution because the resolution criteria states "The market resolves to the average of valid votes in the poll as of 1 day after market close."
50%. As a bettor, I take a very similar approach as @MichaelWheatley said earlier: If there's a clear route to both YES and NO, then I usually assume that the close date is not relevant to the resolution. Example: "Will Musk's deal to buy Twitter close?" has a clear YES case (the deal closing), and a reasonably clear NO case (both sides agree that the deal is off), so without further clarification I'd interpret this without reference to the close date. If there's no obvious route to a NO, I assume the close date is providing the basis for a NO resolution. E.g. "Will Manifold dollars be redeemable for USD?" could eventually happen even if they don't think it's a good idea now, so the best interpretation is whether this will happen by the close date. Of course, it's better for the author to say that explicitly, as there are rare markets that are exceptions to this.
As an author, currently I always explicitly write any date cutoffs in the question text (e.g. "Will numeric markets return by July 8?") or refer to the close date (although because the close date can be updated and the original close date is not shown anywhere, I have realized that I should try to always explicitly write the date).
And if my question is about something without a specific date cutoff, then I simply write the question exactly how I mean it, e.g. "If I take Paxlovid, will I have a Covid rebound?" The timeframe is implicitly based on how long it usually takes for a Covid rebound to develop (which I did not know when opening the question, and researched later), and I can and did adjust the expected timeframe based on how my illness was progressing. I definitely did not mean "Will I have a Covid rebound by the market close date?" (if I had, I would have written it that way as e.g. "Will I have a Covid rebound by 7/10?").
Another case is self-referential markets (e.g. "How much will be bet on this market?") - for these, I think there's a consensus to assume that it means by the close date, and that the close date will not change.
As an author, here's what I've decided to do to avoid ambiguity for questions about whether an event will happen:
- If my question is intended to have a date cutoff, write it explicitly, e.g. "Will numeric markets return by July 8?" I was already doing this because it's just a good practice in general - it also makes the market much easier to interpret when a reader is glancing at the title. If the date cutoff is a minor detail that isn't that important I put it in the question description and leave it out of the title.
- If my question is not intended to have a date cutoff, a good general practice is to specify at what time it would resolve YES/NO e.g. "Resolves YES when the Twitter deal closes. Resolves NO when it becomes clear that the deal will not close, e.g. both parties reach an agreement involving a breakup fee." In addition, I might write something like "The close date is not relevant to the resolution of this market" just to make sure nobody misinterprets it.
(Of course, I might forget sometimes, and other authors have different habits.)
+1 on an optional close date. My default interpretation is that the close date is not relevant, but that requires the question's timeframe to be repeated in the title; I think it would better to make the timeframe more prominent and optional.
I've made several markets with irrelevant close dates that I intend to push out as long as needed.
As a hesitant speculation, I wonder if there's a use for specially defined points in time that are not given as a date, but as an event that someone can trigger. So that when, say, a whole bunch of markets refer to the "end" of a work of fiction, they can share the definition of that, and then one person (a dedicated fan or the author) can close all of them simultaneously and without variation in how they deal with ambiguities (epilogues, omake, deleted scenes...).
Interesting. I've been thinking that something like "criteria date" is better than "close date", because sometimes you want trading to continue a bit after the criteria happens (or vice versa, you sometimes want to end trading ahead of time as the market creator because you're no longer interested in subsidizing the question).
Changing "is the market open" feels somewhat reasonable as the liquidity provider; changing "the day on which the criteria will be evaluated" is somewhat less reasonable, and anyhow it's good to have this distinction.
@Austin Yes, I agree with this, the date used for resolution purposes should be a separate concept from the date that trading closes.
[YES] Unless specified otherwise in the market description/comments, I always interpret "Will X happen?" as "Will X happen before the close date (and will I resolve this market correctly)?" Creators moving the close date without making it clear that they intend to do so reads as changing the terms of the question.
YES
It's just as easy for the market creator to set a close date of "twenty years from now" as "three months from now". If they chose a date that isn't super far in the future, I presume that it was intentional, meaning that it will resolve NO if the event has not happened on that date.
For questions of the form "will X happen" where X can never be fully ruled out (like the SWE question and the abandoned question - it's possible that the engineer quits to work at Manifold in 2050, or that Manifold makes that resolution change in 2050), you need a close date for it to be possible for it to ever resolve NO. I think "'will X happen' questions always resolve NO if X has not happened by the close date" is a better convention than "the close date may or may not be meaningful, depending on the question creator's whims" (though of course "the question creator explicitly states the behavior in the question text" is better still).
@Duncan There was some discussion of this in the Discord https://discord.com/channels/915138780216823849/938171760237477998/987112000112951316 - I'm not convinced that there's actually any benefit to the distinction, I think I would be in favor of removing the "markets can be closed for trading before resolving" functionality entirely.
NO
Looking at each of the examples, I find I assume it will stay unresolved until it happens or the creator becomes reasonably convinced it won't happen. In most cases, there's a reasonable time bound. (One doesn't have a paxlovid rebound 2 years after the initial infection.)
The most ambiguous one to me is the manifold automatic resolution one, since that could in theory happen at any point in the future. I find I assume there's an internal discussion happening that will come to at least a temporary conclusion at some point.
@PeterBorah I do agree with you that these examples mostly look like NO. I feel like I remember many other examples that were clearer YESes (e.g. questions about whether a Manifold feature would be implemented), but I didn't find more examples other than the abandoned question one.