this question has bothered me for a while, so i figured i should put down all my thoughts in one place and make sure they make sense. here are some of my objections: * the question is only meaningful so long as the metaculus community exists. most of our questions are not so; for example in [sustainable off-world presence by 2100](, metaculus could fold in 2025, with some historians/archaeologists noticing the data in 2060, and think "how interesting! ...
@(PeterWildeford) yes, if russian activity occurs only within those territories in [this wikipedia page](, the question will not resolve positively. in my own view, i think this is unlikely, as it appears that there have already been some cyber attacks in ukraine (whether the P5 cite this and whether it's demonstrably proven to be directed by russia is another possible wrinkle, but 2023 is long, and if the invasion starts within the donetsk/luhansk, it could easily spill beyond th...

Release date is set to the date on which the book can be purchased at

this is confusing, as it can mean "date at which pre-orders are announced", which can often be 6 months in advance or more. why not just the obvious and unambiguous "date at which the book is released in at least one country"?

there needs to be a way to delete questions that are drafts or pending mod approval

i think the headline question is a little misleading and click-baity. i would go with "will [english-speaking] twitter have happier [average] content in 2021 than 2020?"

"average" i'm not sure is necessary in the title, or if so, what's the natural way to phrase it. "english-speaking" makes sense but makes the title a little over-wordy

— edited by casens

climate targets are referred to officially as Nationally Determined Contributions. the paris agreement was drafted in Dec 2015 (during COP21), signed in Apr 2016, and put into effect on Nov 4 2016. i've taken [the UN's registry of NDCs]( and [spreadsheeted them]( 94 / 194 parties (48%) submitted an NDC before 20...

@Jgalt the wallets exist, so either someone(s) has the keys, or they don't. it seems like the question is only asking if an authentic transaction out of those wallets ever occurs.

i love the new CDF display on numerical predictions. it would be great if i could have even more specific user input, ie. instead of manually dragging the sliders around, it would be nice if i could enter in some points, like: "5% chance of outcome <X, 33% chance of outcome <Y, 62% chance of outcome <Z..." and then have a function try to match the curve (linearly?) to fit. in my imagination the user would be able to set arbitrary numbers of points, at any percentile.

@DanielFilan "seems gameable" -- you're expecting metaculus users to intentionally get COVID-19 for the MIPs? or for users to refuse to report infection for said MIPS?

second one's a bit risky. if you predict at 1% and vow to not report your own infection, you're at risk of someone else reporting their own infection. first strategy is risky but at least would make a good story about how much people value their health/MIPs

a reminder to anyone putting a huge liklihood on >2100: if there is no AI catastrophe before 2100, this question resolves ambiguous.

— edited by casens

[Reuters: EU chief promises speeded up process for Ukraine to seek membership]( >KYIV, April 8 (Reuters) - European Commission President Ursula von der Leyen pledged on Friday to offer Ukrainian President Volodymyr Zelenskiy a speedier start to Ukraine's bid to become a member of the European Union. >Handing Zelenskiy a questionnaire which will form a starting point for the EU to decide on membership for Kyiv, she said: "It will not as usual be...

@j.m. a third factor: the users who have had a positive result but are choosing not to report for this question (for points, or whatever reason).

from the start of the pandemic, i was suspecting metaculus users to be more cautious and likely to self-isolate than the population average. all these factors, plus the fact that 10% odds come true 10% of the time, don't seem that crazy to me.

@(exmateriae) generally speaking, metaculus is designed as much as possible to encourage you to 1. participate rather than not ([all things equal](, and 2. predict your true belief (rather than try to game the system). so, hopefully it would have been more advantageous for you to predict before the major update, but if the update is what causes your first encounter with the question, there should be nothing wrong with adding a prediction (if it still meets principles 1 and 2). if there's ever a situation w...

i think we can afford to be less indirect with the headline question:

When will the US achieve racial unemployment parity between racial groups labeled 'black' and 'white' by the Bureau of Labor Statistics?

i would change to

When will the US achieve racial unemployment parity between black and white people according to the Bureau of Labor Statistics?

particularly given that the source is in the headline, so the fact that we are deferring to their categorization is implied.

— edited by casens

@Jgalt "some sort of revolution by 2027" ... "if you think 2020 is weird, just wait for 2024. you ain't seen nothing yet"

wow, such falisibility. am very impress

@OpenSystem prior to the UN page updating, "border crossings" was the main statistic they reported, with individual refugees and returns not as clearly communicated. additionally, the resolution criteria says "refugee arrivals from Ukraine", not "refugees arriving from Ukraine", which refers to the crossings themselves, not the refugees. i've made a few edits to clarify this.

this isn't really a relevant comment on any particular question, but it's fascinating to see how concerned i was about COVID-19, documented through my prediction history. i just wanted to express some gratitude, and when the time is right i'll be making a donation on patreon. thanks, anthony & mods.

@SJE median predictions on this question:

feb 15: 1.25 million

mar 15: 330 million

apr 15 (projected): 38 billion

troubling implications...

— edited by casens

@predictors several good comments have been posted raising doubt about the CDC's NowCast, and i find their arguments plausible. i've decided to un-resolve the question to get further input from the community. furthermore, it's been correctly pointed out (by @kievalet ) that the resolution criteria originally called for the question to resolve 1 week *after* the estimate was published by the nowcast. i think it's reasonable to doubt the estimates of the previous 2 weeks of the nowcast, since the CDC themselves admit they are a projection from samples 2...

@Rexracer63 @johnnycaffeine the scoring rules are designed as much as possible for you to predict your true belief, whether the question resolves early or not. you can read the math being worked out here.