World Draughts Forum

Michel, nice work!
My next question would be, could a neural network encode this information (example 5x5), with 99% accuracy?
Or sufficient accuracy to win complicated endgames, or to draw endgames where you have an enormous problem.

Bert

I don't think so. For these endgames, i think many wins ...

After a discussion with Bert, i decided to do a bit of work on the endgame databases again. I increased the generation speed manyfold, and also generated a partial 5 man vs 5 man endgame database. (win/draw/lose only, international draughts)

First i generated the full 1-8 piece endings (120 hours ...

Krzysztof Grzelak wrote: Fri Sep 25, 2026 19:05 Hi Michel,
I'm wondering about this question. Why did you never think about using better compression to reduce the database size?

Krzysztof

That required a lot of research. The problem is the balance between speed and memory amount. I think there will be not too much to win.

I think the premise for mcts is wrong here.

My idea is that the default mcts algorithm expects that the scores follow a normal distribution, but here that is not the case.

I suspect mcts may work if you have a good estimator of the error distribution. And this will need to depend on the position ...

While we are on the subject, i compiled a list of the most frequently accessed databases in dragon's gameplay, with 3 man+1 king vs 3 man+1 king leading the list with 12.8%.

3131 0.12784851 0.400070
3121 0.10273220 0.687230
2131 0.08856994 0.372534
4040 0.05987036 0.973838
4121 0.04104041 0.687221 ...

Version 5.0.1 is now available

It fixes the "missing weights, poor performance" message. In retrospect it may have affected strength in some cases after all.

5.0.1 is much stronger than 5.0. In my internal benchmark the score moved from 42.5% win to 47.2% wins against scan 3.1. A yet to be ...

Interesting. What would be the total execution time for the 8 piece endings?

One tip on compression: dragon does not store the position that are not quiet, instead it looks that up by first performing the capture.

If you add that, you can replace the values of all capture positions by whatever ...

I think Rein's estimate of doubling every 18 weeks is fairly good on the spot.

The current AI compared to that in februari this year is already incredable. Recursive self improvement is on the horizon in my opinion, where AI will write the next generation of AI's. (and do that in a loop)

Only ...

Out of curiosity I let Codex design a neural network implementation with exactly the same parameters that the Ares network uses.

First of all, the code generated by Codex contains a lot of unnecessary crap that will only hurt performance. Secondly, it was not able to generate code that comes ...

Hi Michel,

What weights to use please?
I launched et dxp match and so "missing weights, poor performance"

Friendly, Sidiki

Hi Sidiki,

You can ignore the warnings. They are about the old evaluation function, but dragon does not use them any more. I will remove the messages in a new version ...

Michel, thanks for sharing.
What is ELO wise still the gap compared with KR, with 1 core 1 min games, or a similar setup?

Bert


Against kingsrow, dragon is losing about one in 6 games, for fisher time 1 sec+0.1 sec, single core.

I intent do tests against scan 3.1 only, as that is easier to ...

You can download version 5.0: https://www.mdgsoftware.nl/draughts/index.html

* New, stronger evaluation function
* fixed bug for thread selection on 14, 18 core machines
* split between for cpu's SSE2 and AVX2 support (avx2) avx2 is +8% faster
* Installer now uses PGO version for +5% speed ...

Dragon can, but you need to put a pdn file with the positions in the dxp directory. I forgot to add that to the installer.

I will fix in the next update, working on that.

You will also be able to run damexchange directly from the engine, for any list of pdn files. No GUI overhead. Just start the ...

I often let the engine play games of 10 sec. with 0.1 sec. increment, usually that doesn't give problems. The only problem that arises is when the program starts using endgame databases, then it will suddenly take a long time to load and decompress them, only a pre-loaded 6P EGDB (or less) will ...

How about a match at 1 second+0.1 second per move? With incremental time control you also don't need to stop at 75 moves, just let the engine run until the 25 move rule forces a draw.

I am up for it. It would take some modifications for me though. Currently the GUI would eat up the thinking time ...