For what it's worth, from my point of view this should resolve negatively. The scenario we're currently in is, to me, exactly what I would have imagined negative resolution to this question would have looked like, prior to the election.

I'd rather not have a precedent set for resolving questions in ways contrary to how the resolution is described in the question text. I know this one seems pretty harmless, but the norm of always resolving questions as predictably as possible is a good one to keep.

I think the moral of this story is instead, "why did we specify ambiguous resolution so aggressively if, in the event of that happening, we'd be upset and want to change our minds?"

@(tenthkrige) This is possible if, for example, you spent some of the question's history predicting this was very unlikely (eg. 1%), and some of the question's history predicting that this was very likely (eg. 99%). Because of the log-scoring component of the points you receive, you are punished very strongly for being confident and wrong - more strongly than you are rewarded for being confident and correct. So it might be that regardless of the resolution you are getting some large negative score from the period of time for which you were confident and...

I see comments every now and then on the SpaceX questions referring to a constant I call the "Musk multiplier" (let's call it m). It is defined as follows: when Elon Musk predicts that event X will happen after time t has elapsed, the event will actually occur after time t*m has elapsed.

It is generally assumed that m>1.

It would take some operationalisation but predicting a value for m would be fun.

I think the problem here is that the early-resolution score reduction is appropriate for questions of the form "Will X happen before Y date" - otherwise there's an incentive to over-estimate the likelihood of a positive resolution since a negative resolution can't happen early, only a positive resolution can. But for questions like this one, the early close does not contain any information about the resolution of the question, and so the early-resolution score reduction is inappropriate. So I think there's a need to tag these two kinds of questions, the...

Presumably the question is not asking whether or not at least one large attack is prevented (which would be very hard to know), but whether any such large attacks successfully occur.

@traviswfisher - I think the other big consideration here is that, plausibly, lots of the people who die from this Corona Virus might have otherwise died this year anyway and so are already priced in to the overall population growth stats. For what it's worth I'm still at 99%.

@Reprisal - I think in some cases I would have sympathy with this argument. But in this case, as far as I can tell, the question links to previous month's numbers, and they (ie. April) contain this same [erroneous methodology/anomalous definition/difference in calculation]. So a reasonable person might well have expected May to be calculated in the same way as April was, and that's what the question has done. I favour keeping the resolution as is, using the official (if unsurprising) statistics.

I think the repeated references to the `current coalition' strongly imply that a continuation of the status quo would resolve positively. Given that the coalition in the legislature has continued and that, as is pointed out below, the previous commission also contained some non-coalition members, a positive resolution seems right here. Perhaps comments under the question can be used to clarify genuinely ambiguous cases - but they shouldn't be retroactively changing what seem to be pretty clear resolution terms. (Usual caveats apply - I predicted 60% f...

@converse - If we get an out-of-office reply it still resolves positive though

@(Jgalt) My view is also that this should resolve as 202. Some results totals have made the error of counting Hoyle in the Labour numbers, presumably because they just checked who retained their seats. If you search for Hoyle's constituency, Chorley, [on the BBC results page]( you will see it is coloured in grey and listed as SPK not LAB. But in their totals SPK is not present and he is erroneously listed as LAB. Since the BBC results totals say 203 but the map (correctly) has 202 I'd prefer going 202 but ...

It now seems confirmed that the UK government and the EU have agreed an extension until the 31st of October 2019 (for a spooky Halloween Brexit) and so I think we can expect this question to resolve negatively tomorrow evening.

PS: Given that the last two deadlines (March 29th, April 12th) have been extended, I think we should have a question as to whether this one gets extended as well.

— edited by James

Do mice which are cryogenically frozen and then resuscitated (or similar) count here?

Betfair's current price of ~1.33 suggests a ~75% chance of repeal winning.

@tenthkrige @trevorandersen - I think that we can just multiply the two survivor rates because the 118->120 rate is only measured for the people who survive to 118. So even if the people who survive from 116->118 are in some sense more likely to survive to 120 than expected, the same bias was also present in the population that was used to generate the 118-120 rate and so is already incorporated into that figure.

I would suggest resolving as ambiguous - and then we've all learned the lesson that questions about whether a study finds statistical significance needs to specify which subdivisions of the population count and which don't.

Without getting all 'if a tree falls and nobody hears it...' - I do feel that 'additional significant evidence emerging' does require that evidence to be possible to find in a reasonable amount of time - and given that nobody has been able to do it in almost a year now suggests that this ought to resolve negatively.

@Jotto I'm concerned about the detrimental effects on both my Metaculus score and the planet if this resolves positively.

Regarding early resolutions and similar: I'm generally of the opinion that `the game' is predicting things some time in advance of the thing happening (or not happening), so as to test how good we are at predicting unknown, future events. In cases where we know when resolution is likely to happen (ie. `will X event on Y date be a success or failure') we should close the question well in advance of the event. This also means that when the possible resolution date is spread across a known region of time (eg. `will X team win tournament Y') - we should c...