@AABoyles Thank you! I love your hacks! :)

@Jgalt We might see a reference to a third derivative (change in (change in) job growth), perhaps not seen since Nixon!

@silly I think he might find a way to leave office without explicitly conceding. Maybe he just won’t say anything. I agree it’s very unlikely (< 5%) he continues to claim to be president after Inauguration Day if he loses the election (whatever “losing the election” means...). Or maybe he’ll say something like “The election was fraudulent. I am not conceding. But I will leave office because I am being illegally forced out.”

— edited by Cory

@(krtnu) I’m not sure when TradingEconomics first reported the forecast, but I updated my prediction in the 24 hours before closure based on forecasts of 850k. [This September 27 Yahoo Finance article](https://finance.yahoo.com/news/september-jobs-report-first-presidential-debate-what-to-know-in-the-week-ahead-171840680.html), for example, cites an 850k estimate from Bloomberg data. The TradingEconomics page says both that “market forecasts” were 850k (actual 661k), and that private payroll forecasts were 850k (actual 877k). Interestingly, the differenc...

In order to view my past prediction PDFs, I've created this spreadsheet. Predictors can create a new copy and paste into row 2 the target continuous prediction from this hack to download predictions courtesy @AABoyles. Of course feel free to improve it; I don't have any spreadsheet experience.

I'm at 85%. I think the probability of the pre-conventions 538 model giving Trump more than The Economist's current 8% is around 95%. Most of my uncertainty is in the Economist adjusting their model to include more uncertainty before the 538 model is released. On July 20, [G. Elliott Morris said](https://twitter.com/gelliottmorris/status/1285332756671672324) they want to increase hedging against extreme outcomes in their model, and "we are talking about aggregate probabilistic changes around 5 points at max." On July 18, Nate and Elliott had a short [di...

@Tamay Well that didn't take long! I think this can resolve positively now, currently trading at over a $2.06T market cap in 2018 dollars (based on Q2's price deflation).

We're getting closer! For the purposes of these questions, when would a multi-dose vaccine be considered "administered"? For these questions that require a certain number of people to have been vaccinated, must a single vaccine have been administered the required number of times, or can multiple vaccines contribute to the number? I assume that when an "article" is required for resolution, and "pre-prints" are also mentioned in the resolution criteria (so this does not include the number-of-people-vaccinated questions that I assume may resolve based on...
@(Bernhard) Eyeballing the [accompanying chart (figure 6)](https://www.federalreserve.gov/images/powell-figure6-20210210.png) from [Powell's speech](https://www.federalreserve.gov/newsevents/speech/powell20210210a.htm), it looks like the misclassification error makes up around 0.7 percentage points of that ~3.7 percentage point difference. This is a really useful visualization of how the unemployment rate methodology might differ from intuition as people technically drop out of the labor force. I don't know how quickly I should expect this error to be c...
@(kievalet) Metaculus uses a [specific definition of "credible source":](https://www.metaculus.com/help/faq/#definitions) >A "credible source" will be taken to be an online or in-print published story from a journalistic source, or information publicly posted on a the website of an organization by that organization making public information pertaining to that organization, or in another source where the preponderance of evidence suggests that the information is correct and that there is no significant controversy surrounding the information or its corre...
@(BogdanIonutCirstea) Cool! I think [Google measures their TPU performance in bfloat16 operations](https://www.nextplatform.com/2018/05/10/tearing-apart-googles-tpu-3-0-ai-coprocessor/), and their FP16 performance is significantly worse. But bfloat operations should probably still count as computations for this question. More importantly, I can't confirm that a TPUv3 core can do 90+ teraflops; I think it might actually be half that. [Google advertises a "Cloud TPU v3" product with 420 teraflops of performance](https://cloud.google.com/tpu/), and I think...
@(yshemesh) I'm confused about the use of a tie-breaker for this lottery. Why is it a problem for multiple participants to have the same odds of winning a prize? What impact will the tie-breaker have on a participant's odds? I'm also confused about why questions 11-14 are labeled "tie-breakers." I understand that these questions will not have a rationale component and will only be used for brier accuracy and categorical accuracy scores, but I don't see how they are used as tie-breakers. It looks like when a tie-breaker is needed, final completion time i...
@(Jgalt) The question specifies that the system must have actually delivered the payload at the stated price. According to [Wikipedia's list of Falcon 9 launches](https://en.wikipedia.org/wiki/List_of_Falcon_9_and_Falcon_Heavy_launches), besides the internal Starlink missions (15,600 kg) and Crew Dragon missions (~12,000 kg) for NASA (which probably has a unique contract and pays more per launch than other clients), the most massive payloads launched are 9,600 kg. Assuming this cost $50 million, does that mean that right now the lowest cost is probably $...

@yshemesh Thanks for putting this on; I'm really excited about the unprecedented prize pool and to see the results and my personal rationale feedback! I just want to let you know that I plan on waiting until the deadline to submit my predictions because I think that is the best way to maximize my expected winnings. The questions should resolve relatively soon, so I think the extra week of data and contemplation should be worth the reduced odds of winning a submission-time tiebreaker.

@casens Is the main difference between this question and this very similar question about US COVID-19 deaths in 2022 that this question prioritizes CDC data, and the other question prioritizes WHO data?

@(j.m.) I think this is a tough one. If "US county level" was mostly meant to exclude L4 geofence stuff, then I would feel better about this qualifying. On the other hand, I would be surprised if this Tesla FSD beta is enabled for more than 1000 drivers, and that doesn't feel satisfyingingly qualifying to me. On busy city streets, I still think [a human would intervene 2+ times per mile](https://www.youtube.com/watch?v=VosewaoYoIg). My final prediction was 20%, and I am pretty surprised and very impressed by FSD this year. When this closed on June 1, I p...
@admins Looks like this is going to resolve positively for the week of June 14-20, with 4203 deaths after the previous week's 5277. The COVID Tracking Project has [a nice chart of daily deaths over time](https://twitter.com/COVID19Tracking/status/1274473854849609728). There's some weirdness with "this question will resolve using the data on the "US Daily 4pm ET" tab on 2020-06-015 at 16:00 EST," but I think it's obvious what was meant. There's also some weirdness in [the expert survey](https://works.bepress.com/mcandrew/8/), which when read literally wo...

@Matthew_Barnett Yeah, I don't get it. Before, I would have predicted $2-5 million (50%). Wonder who the $20MM+ bidders were. I still think in 5 years, or (likely earlier?) when 100+ tourists have been to space, the price will be well under $500k. The value of being among the first private space tourists, on Blue Origin, with Jeff Bezos, etc. really surprised me. Maybe once you have such a fortune, such scarce opportunities become more valuable than I expected.

— edited by Cory