Currently the answers in a multiple choice question are mutually exclusive, they must sum to 100%. There are many questions where I'd like to have them either sum to a different number (e.g. "Which movies will be nominated for Oscars Best Picture" where 10 answers will be selected) or not be forced to sum to a specific number at all (e.g. "Will Starship next launch before <August/September/October/November>?" or "Which countries will join NATO this year?"). This question is about the former.
Resolves YES if there's an option for answers in a multiple choice market to add to a fixed number other than 1 by the date in the title (even if it is later removed). It must be supported in the Manifold interface for regular users (API support alone does not count, admin-only support does not count).
If the market allows each answer to be completely independent, that does not qualify, since it does not sum to a fixed answer other than 1. (This would be amazing as well, but it doesn't give the automatic arbitrage / market-making that you get when it sums to a fixed number, so it's different.)
This is one of the features that Manifold has said they plan for multiple choice:
Adding new answers (Done!)
Sell button (Done!)
An option for independent answers that don't sum to 100%, and/or answers that sum to something other than 100%.
Related questions
🏅 Top traders
# | Name | Total profit |
---|---|---|
1 | Ṁ710 | |
2 | Ṁ461 | |
3 | Ṁ112 | |
4 | Ṁ99 | |
5 | Ṁ30 |
e.g. "Which movies will be nominated for Oscars Best Picture" where 10 answers will be selected
...
automatic arbitrage / market-making that you get when it sums to a fixed number
For best market making / arbitrage, I think you also really want the ability to limit the range that a specific answer can pay out as. For the Oscars example, 10 answers will pay out at 1, the rest at zero, and the total payout sums to 10. If movie X is trading at 0.5, that represents a 50% chance it will be nominated. Betting it up or down should be symmetric in terms of shares per M spent; it should take M 0.5/share to buy either yes or no. But without the defined range of 0-1 per option (in addition to the defined sum of 10), buying yes on that market will cost 0.5/share and buying no will cost 9.5/share.
@EvanDaniel Actually, the case of best picture is quite interesting. From my recollection, there can be up to 10 nominees, but there are often only eight or nine. The idea that the number needs to be fixed in advance makes this a little bit more tricky. I would personally have much more use for an ability to specify a narrow range. One way around this is to allocate points for "unused" slots to a default answer, but it's a little icky.
@Eliza Yeah, it's a bit icky, but good naming might help? You name one slot "there will be 9 or fewer nominees", another "there will be 8 or fewer", etc. as needed.
Can you update the description based on the comments below?
It currently does not say that the unlinked kind is not sufficient for a Yes resolution. It just says all you care about is probability not equaling one. But I'm pretty sure you mean to specifically exclude the case where answers have completely unlinked probability.
@Eliza I inserted the word "fixed" in the text "number other than 1". That should clear it up right?
I think with the context of the first paragraph the other interpretation is clearly ruled out, but I always prefer the resolution criteria to be clear on their own, so sorry to anyone who was confused about that.
@Eliza The unlinked multiple choice variant should satisfy this from the market description:
> not be forced to sum to a specific number at all
@Eliza Wait, are you saying that part doesn't count? Why did you put it in the description?
@Joshua I don't really see how the independent one would not count for this. They would sum to something other than 100%, trivially. Nothing in the description says that this would not count?
@Eliza My understanding is it has to add to a specific number that’s not 1, not just be unlinked. The key line in the description is “This question is about the former.”
I agree it’s confusing, I also got confused below
@EliLifland I read that part too and he quoted the exact same part as me. I thought he was saying it was very clear.
@Eliza He said “Yes” to my question about whether the markets are different, then saying there are two separate things in the quoted text; this question requires, the first, second question requires the latter.
@Eliza Sorry for the confusion. When I wrote "number other than 1" I meant a fixed number, but I agree that isn't clear, edited.
Is there a difference between this and https://manifold.markets/jack/will-manifold-add-multiple-choice-w-41b1cfa25c91
@EliLifland Yes. Two separate things:
There are many questions where I'd like to have them either sum to a different number (e.g. "Which movies will be nominated for Oscars Best Picture" where 10 answers will be selected) or not be forced to sum to a specific number at all (e.g. "Will Starship next launch before <August/September/October/November>?" or "Which countries will join NATO this year?").
See the resolution criteria which I believe are sufficiently clear, if there are questions lmk.
@EliLifland I assume independent markets on each answer wouldn't count because they don't sum to a fixed number? At least that's one way this market would be different.
@jack I don't think the resolution criteria here are clear when read literally and without contrasting with your other market, since an unlinked market would frequently "add to a number other than one". I would prefer "fixed number other than one".