See https://www.lesswrong.com/posts/KTEciTeFwL2tTujZk/lw-petrov-day-2022-monday-9-26
This year the pool of users who can use the button expands over time.
Resolves YES if the frontpage is never taken down, otherwise resolves PROB to the percentage of time elapsed between the start of the event (3am UTC) and 24 hours later (3am UTC on the 27th) at the time the frontpage is taken down.
Original market:
Some LessWrong discussion of this market: https://www.lesswrong.com/posts/DpDnXHcPejd9tn8R5/ambiguity-in-prediction-market-resolution-is-harmful
I've decided to go with N/A resolution.
I do think it is a strength of Manifold that markets can be resolved according to the commonsense interpretation of a question even in the face of annoying technicalities, but in this case:
-The issue here wasn't just an annoying technicality, but something where several people had reasonable arguments that it was a straightforward and obvious resolution.
-I did not adequately clarify in advance how I would deal with what I perceived as ambiguous situations.
-Multiple people suggested resolving N/A, or said they would consider resolving to 80% dishonorable.
I apologize for the insufficiently detailed market criteria. I have added more details to some of my existing markets, and in the future I will lean more toward rules-as-written or N/A resolution in case of ambiguity.
Happy Petrov Day
According to habryka, the site went down at "~5:40pm PT", which is 12:40am UTC, or 9 2/3 hours out of 12, or about 80.6%.
What minimum karma does this correspond to? If my math is correct, 80.6% corresponds to a little over 19 hours into the challenge, which I believe means anyone with 400 or more karma could have pressed it. But I thought I remembered seeing it go down to 300, so I'm worried I've made a mistake somewhere.
@IsaacKing Nope, my math was just flat-out wrong here, I divided out of 12 hours instead of 24. Correct percentage is about 90.3. Good thing this market resolved N/A haha
Having read lots of comments here and elsewhere I am currently conflicted about whether to resolve N/A or not. If I don't resolve N/A, I will resolve in accordance with my previous ruling (the first one didn't count), so if you want to keep trading you can trade based on that premise.
@Multicore I think there are good arguments that 10% was the correct resolution, but I don't think it's an option after I ruled against it.
It's difficult to incentivize people to not press the button, but here's an attempt: If we successfully get through Petrov day without anyone pressing the button (other than the person who has already done so via the bug), I will donate $50 USD to a charity selected by majority vote.
Feelsadman. I'm pretty sure the person who pressed it initially had enough karma.
@mkualquiera There was a bit of a clarifying edit after habryka initially commented; it currently says that due to the coding error "anyone with zero karma (but only exactly zero karma)" was able to launch the missiles, which to me seems to imply that users with more than zero karma could not launch them.
@ev Again, I don't think that's correct. The site went down at exactly 6:01 UCT, which to me is a very strong signal that someone with less than 2100 karma was waiting eagerly to launch the missile, and did so. In fact, I personally know one such person (unfortunately they're asleep right now).
Furthermore, I have an account with 0 karma and I was personally able to verify that I was not able to launch the missile.
Well the site WAS taken down with someone pressing the red button. But apparently it was due to a coding error allowing 0 karma users to press it and the site went back up.
@DavidChee That should still resolve the market. This is a risk they are taking by deliberately adding a security vulnerability to the site.
@DavidChee I will rule that the false alarm did not count for the purposes of this market.
@Multicore Another confusion here is that the code wasn't posted until 4:05am UTC, an hour after the intended start time (which was 3AM UTC). But it looks like the button itself was displayed at 3AM UTC? (It's currently 12:30 BST and the karma threshold was 1500. So it would have been 1500 starting at 12:00 BST, and 2300 starting at 04:00 BST, which was 03:00 UTC.)
@Multicore looks like there's a post on lw disagreeing with this ruling. I personally can see it either way.
@Multicore Have you considered N/A resolution? I think in cases where one thinks the spirit of the question and the text of the question would lead to a different answer, N/A is probably best.
I think there's reasonable disagreement about the spirit of the question, but the letter of the question is very clear.
@Multicore Just to personally weigh in I lean slightly more towards resolving to 10% (or wtv it was when it went down even if because of a bug). As what occurred exactly matches the resolution criteria is described as even if it was not what was intended.
But I also think resolving N/A is very reasonable as an unexpected variable was introduced which causes the resolution description to no longer align with what was clearly the intention of the market.
The decision is ultimately yours though. Good luck making it haha
There's at least one high-karma person who claims ~50% chance of pressing the red button at strictly less than 25% of the way through the day. And that's just one where the discussion happened to make its way into the below Manifold market. There are probably multiple other similar people who kept their thinking private.
