noobpwnftw wrote: ↑Fri Mar 17, 2023 6:26 pm
They'll probably just use it to make more Mittens. That said, it looks like a good deal on your end.
Well, they already had the right to do this (and apparently did with Mittens), so if that was their sole intention, they wasted some significant money. I should mention that we (meaning Mark and I) will get a sizable bonus if Komodo Dragon does manage to pass Stockfish by specified criteria, so we still have financial incentive to help improve Komodo.
That is, there is hope... just for what ?! Hmm... probably for new bots and hundreds more videos on YT with kittens, doggies and frogs as the new increasingly powerful (Komodo Dragon) Chess.com bots.
I'd like to be wrong, but looking at the angle of the Mittens bot and what you wrote Larry, I don't have rather high hopes, unfortunately....
Do remember that Mittens is/was more popular than any of us (or any engine...outside perhaps Stockfish) will ever be. Strong is strong...use of a strong engine is where it's at.
Popularity is not the only factor that matters in chess or in other fields.
Popularity is like a candle flame - it can grow quickly, but die out just as quickly
noobpwnftw wrote: ↑Fri Mar 17, 2023 6:26 pm
They'll probably just use it to make more Mittens. That said, it looks like a good deal on your end.
Well, they already had the right to do this (and apparently did with Mittens), so if that was their sole intention, they wasted some significant money. I should mention that we (meaning Mark and I) will get a sizable bonus if Komodo Dragon does manage to pass Stockfish by specified criteria, so we still have financial incentive to help improve Komodo.
That is, there is hope... just for what ?! Hmm... probably for new bots and hundreds more videos on YT with kittens, doggies and frogs as the new increasingly powerful (Komodo Dragon) Chess.com bots.
I'd like to be wrong, but looking at the angle of the Mittens bot and what you wrote Larry, I don't have rather high hopes, unfortunately....
Do remember that Mittens is/was more popular than any of us (or any engine...outside perhaps Stockfish) will ever be. Strong is strong...use of a strong engine is where it's at.
Popularity is not the only factor that matters in chess or in other fields.
Popularity is like a candle flame - it can grow quickly, but die out just as quickly
Of course it is not. But single minded pursuit of (since we are talking engines) 'strength' in chess engines is ultimately a dead end. It's what people can do with the things that matters most. Nothing wrong with using one to power a trash talking feline. May your life be filled with as much joy as others find in the things you seem to sneer at.
lkaufman wrote: ↑Fri Mar 17, 2023 3:57 pm
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 mentioned this to you a while ago when NNUE first came out Enjoy retirement though!
With the immense capability of recent large language models, which are closing on general AI, chess with a tiny or big NN holds little interest for me anyway.
lkaufman wrote: ↑Fri Mar 17, 2023 3:57 pm
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 mentioned this to you a while ago when NNUE first came out Enjoy retirement though!
With the immense capability of recent large language models, which are closing on general AI, chess with a tiny or big NN holds little interest for me anyway.
Yes, I did rather expect this, but I thought at the time that NNUE would mean drastic revisions to the engine (at the very least huge parameter changes) would prove beneficial, but it turned out that only modest tweaking of the engine was called for, at least as far as I can tell. Even the MCTS version of Komodo didn't seem to need much tuning with NNUE. I imagine that someone will come up with a whole new approach to best utilize nets in chess, but the fact that it only really matters when starting with unbalanced positions removes a lot of incentive. Maybe the next breakthru will again occur in shogi, where it is still likely that the engines can improve by hundreds of elo without artificial openings.
I'm a bit disappointed by the lack of progress in Komodo, Stockfish and Fritz.
Maybe the lack of progress is because neural networks are becoming an obsolete technology.
In the future we'll have better technologies than neural networks like quantum computers, artifical general intelligence, optical computers etc. Scientists are even working on making an artifial brain.
.