There are no more players without completed games!
The dot on 0-0 coordinates in the Games Performance chart has vanished.
Now @trohde can rightfully claim to be the slowest player in the tournament… well, one of the two slowest players.
@Lys Have you created a best fit that predicts when the first round will be completed? I feel like at this point you should be able to make a pretty accurate prediction given the data.
Progresses are getting slower and slower.
I don’t have a reliable formula to predict tournament behaviour.
I tried simple trend curves (poly and log) but they don’t fit.
Best fit so far is the capacitor charging curve, which isn’t very reassuring: coeff a is the full capacitor charge. According to my curve, we’ll reach 9850 completed games next may and that’s it. If the comparison is correct, we won’t see the end of first round…
We tried also radioactive decay formula, which resembles our data but doesn’t fit very well. Looks like we are slower, we are decreasing less than expected using that curve… and that’s a curve that doesn’t actually reach zero! Does it?
Estimations for worst case (using all the time and vacations and weekends and so on) were talking about 16 years or so… the real answer should be in between those boudaries: next may and 16 years in the future.
That isn’t a very sharp prediction, is it?
Making predictions about the most extreme outlier is a fundamentally difficult problem, since the most extreme outlier is inherently something that does not fit well with the general trend of the rest of the data. In this context, I think the longest game of the first round may likely be due to some extraordinary factors that are almost unique to that particular game. So, the best we can do is make wild guesses when extrapolating the data.
Another approach is to estimate the statistics (like mean and standard deviation) of the length of games finished so far. That would be a biased estimate, since we don’t yet have the data for the roughly 2.5% slowest games. One could try to correct for this by also discarding the 2.5% fastest games.
Given these statistics, one could make a simplifying assumption that the each game duration is a random variable with either a Gaussian or exponential distribution (with each game being independently and identically distributed). Then, estimating the length of the longest game involves looking at the statistics (like mean and confidence intervals) of the maximum of thousands of independent and identical Gaussian or exponential random variables.
Another way to make a prediction: the slowest games are advancing at a pace of 1 move/week. My prediction is that the first round will be finished by the end of 2025.
We’re lightyears away already … when we return and the tournament ends, cats (dogs, cockroaches …) will have evolved to talk, walk upright, and have opposed thumbs …