BREAKING news : chess.com buys Komodo/Dragon !!
chess.com has acquired Komodo/Dragon chess
GM Larry Kaufman quoted below . Besides chess.com owning a strong program ( 2nd behind Stockfish though). Larry believes top engines at classic speed draw games, if they dont play subpar opening. 98% of standard chess solved is his quote. scary !
Five years ago, chess.com purchased a stake in KomodoChess along with rights to use our software. Now the merger is complete; chess.com has bought out KomodoChess entirely. Chess.com will now run the website, pay for further development of Komodo, and for the most part take over the responsibilities of both myself and Mark Lefler. Mark and I will remain as paid consultants thru 2025, but chess.com will make the decisions. Dietrich Kappe remains onboard as the NN trainer. It is not yet clear who will be the main programmer or programmers, but development will continue, although perhaps in different directions. The goal of becoming the world's number one engine will remain as a prime goal, but it is expected that this will take some major work and hence I wouldn't expect any quick results.
There were several reasons for this:
1. I reached the age of 75, pretty old for such a tech-heavy job. Mark is about a decade younger, but even so that's a normal retirement age here (well past it in France, I read!).
2. Progress with NN training came to a halt, I won't pretend to know why. I have confidence in Dietrich to find a solution, but time is not on my side.
3. Progress with the engine also came to a virtual halt after the release of Dragon 3.2. I believe only a major re-write, or some major discovery by a new person, can solve the problem.
4. KomodoChess sales weren't sufficient to hire a new full-time programmer or to finance massively more net training, but chess.com has more incentive than just KomodoChess sales to make Komodo Dragon the best, so that need not deter them.
5. In my opinion, Komodo Dragon and Stockfish have effectively "solved" standard chess in the sense that with long time limits and large thread counts, they are both probably invincible if allowed to choose their own openings or at least get to choose from near-optimal openings. This seems pretty clear from many results; even the CCRL 40/15 4 cpu list shows just tiny elo differences among the top Stockfish and Dragon entities, and that's with some dubious openings included. Of course these engines are far from perfect in playing positions that are not as clearly drawn as the standard start position, there are still large gaps from version to version when tested with unbalanced openings such as we see in TCEC or CCC. But for me, the challenge of making improvements that only show up in these artificial situations is not so exciting as it used to be. Even chess960 is over 98% drawn on big hardware in Rapid, based on CCC event. I still want to help Komodo make it to the top spot, but I don't want to spend the rest of my life on this when the CCRL Rapid gap is down to five Elo points already.
6. The discovery of NNUE was wonderful for both Stockfish and Komodo, but it greatly diminished my own role. I won't pretend to be an expert on NNs; I have of course learned a bit about them during these last couple of years, but not enough to contribute more than a little to the NN side of the project. I think the same applies more or less for Mark.
I hope that I am still able to make some significant contributions to the further development of Komodo Dragon, I remain interested in and involved with computer chess. I think that Mark feels the same
GM LARRY Kaufmann on Talkchess
but Stockfish remains #1 and 100% free ; stockfish.org
https://stockfishchess.org/download/
