Lothar Jung schrieb:
Bei langen TCs ändern sich halt die Stärkerelationen.
Aber geh, was du nicht sagst.
Vor allem auch, weil man länger warten muss, bis genug Partien gespielt sind
Nein, im Ernst, Lc0 hat gegen alle Schwächeren gemeinsam schon 600 solche und SF erst 200, das verzerrt das Bild schon noch ein bisschen, aus den Comments by the chairman:
Zitat:
This last rating list of 2023 consists of five new entrants. First one out is the latest version from Team Stockfish, named Stockfish 16. We have as usual used an opening book from Fauzi Dabat for the testing, and this one is called: "Aggressive 5.6 by fauzi.abk". It's of course still early on in the testing, but Stockfish 16 has reached the very strong rating of 3582 after the first 200 games and have taken over the second place in the rating list.
Even though the earlier Stockfish 15 gained 3 points since the last list, Stockfish 16 now has a gap of 9 rating points ahead of its former version. As Lc0 lost a considerable amount of points since the last rating list, the gap between Stockfish and Lc0 have now gone from 40 to just 4 points. It will surely be exciting to see how the rating will continue to fluctuate in the top of the list as more games will be played with all the top programs.
As incidentally mentioned, Lc0 wasn't able to keep up with its strong start in the last rating list and with more than 300 games added since last time, its rating has been lowered with 24 points. Lc0 still holds the number one spot in the list, but now with a much smaller gap to its nearest competitor - Stockfish.
Lc0 hat ja auch noch eine error bar von ca. 30, da ist natürlich nicht nur im Rating, sondern auch im Ranking noch viel Zufall drin.
Trotzdem, wie gesagt, feine Liste, sie war zeitlich immer schon etwas hinter den anderen her, hoffentlich wird die Irrtumswahrscheinlichkeit noch klein genug, bis SF16 wieder durch SF17 ersetzt wird