I see you’re using the esp() function. That should be great for predicting the future. 
What does a graph of λ = log(N(0)/N(t))/t against t look like? It should be roughly constant if N(t) is exponential.
1 Like
Is that some kind of inverse formula? Why N(0)/N(t)? Shouldn’t it be N(t)/N(0)?
Here I drew both.
On top is λ = log(N(0)/N(t))/t
On bottom is λ = -ln(N(t)/N(0))/t (is that the inverse function for N(t)=N(0)*e^(-λt) ? My math is quite rusty)
Neither one looks very constant.
log(x/y) = -log(y/x)
My log() was meant to be natural log (base e). I’m guessing your ln() is natural and your log() is base 10.
If so, your log() and ln() differ by a constant factor so I don’t know why the two graphs look so different.
1 Like
Nice! I didn’t recall that.
Should be so, yes.
I’ll double check that
My fault!
Time value slipped inside the logarithm. 
After changing LOG() to LN() and moving t to its place, curves are identical.
Much more stable now!
But still not very constant, since a change of 0.0001 in λ is noticeable in the curve.
Very interesting. The curves for disqualifications, resignations, time-outs and wins probably have very different shapes, which may account for what looks like three distinct phases in this latest graph, where they’re all added together. The graph relates to the probability of a given game finishing on a given day but shows the change much more clearly than the previous, non-logarithmic one. The current downward slope means my previous estimated completion date, based on a constant three-month half life, was optimistic.
1 Like
If you’re talking about disqualifications and resignations from the Tournament, I don’t think they’re affecting directly the games shape: resignation happens when you decide to click that button, but doesn’t stop games. I didn’t check, but probably we have resigned players which are still playing their games in the tournament.
Similar situation for disqualified players: they lost one game by timeout but that didn’t stop their other games.
So games are proceeding.
Resigning a game seems to me a normal way to end a game.
Timeouts had some bumps: about three days after tournament start we had massive black’s timeouts. One month later the same happened for white. Those were all players that didn’t play their first move.
Other than that, timeouts are still happening in a quite regular way. You can check that in the disqualified players chart.
We could check if the timeouts rate matches the completed games rate. Maybe timeouts are exponentially decaying too.
1 Like
I don’t want to download each single game, but we could compare completed games and new disqualifications.
Going backwards:
29 vs 4
28 vs 7
61 vs 5
54 vs 9
…
I can’t see a trend, but I could make a chart in next days
1 Like
Bad time for black
1 Like
- 752 games left
- 30 games completed
- 3 players disqualified
- none resigned (it’s the second time in a row)
- 22 players completed all of their games, one of them with 9 victories
- one group completed all of their games
I checked that. I was wrong.
For what I can see, in this tournament’s history there’s only one player that did resign from tournament before completing his games. There are no games in progress at the moment by resigned players.
The opposite for disqualified players: there are now 170 games in progress in which at least one player got disqualified.
Here is a chart for games by disqualified players (at least one of them):
I thought about it.
At first my belief was that new disqualifications could be a proxy for games ended by timeout, but that’s wrong: players that are already disqualified may lose other games by timeout.
So we have to chech each game for that. I’m not planning to do that in a short time.
5 Likes
This meme by @Kosh seems to be a reference to this tournament
It reminded me of this great story by Isaac Asimov, “The Last Question”
https://templatetraining.princeton.edu/sites/training/files/the_last_question_-_issac_asimov.pdf
4 Likes
I simply check the games finished from time to time. And I noticed a trend.
Six games finished since the last time I checked:
1 Like
The first game, Hakmans 5k – nutpen 1k, continued like this in the top right:
If I were playing against an opponent four ranks higher than me, though, I’d push through immediately:
It’s the difference between a simple advantage and a complex one.
2 Likes
1 Like
See why I get the impression timeouts are more prevalent at this point?
1 Like
Large parts of the western world are coming out of various levels of lockdown. Also he northern hemisphere approaches summer (although it doesn’t feel like that here in Hamburg, I can assure you). Might be enough to explain more timeouts.
3 Likes
I think it’s just natural process of a tournament
1 Like