Rein Halbersma wrote:
Congrats on your triple crown! What improvements did you make, and what did you notice in other programs? Will you publish your changes?
I see that the rate of your questions has improved as well
At least the authors of the following engines have claimed progress: Deep Nikita, Dragon Draughts, GWD, Scan, Sjende Blyn, and Slagzet.com. I could clearly see it for GWD and Slagzet.com. A sample of two games looks tiny, but the authors confirmed that they observed a large progress in other games.
Nothing visible for the top programs of course; only long matches will hopefully confirm anything. Elo-wise there are only few changes in Scan; most of the work was re-organising the code around immutable position representations. Eval is better and faster, thanks to larger patterns (which reduces their number) and higher-quality examples. QS is a tiny bit better as well by extending "opponent can capture" positions at the first ply of QS. Nothing else that I recall.
I am planning a 3.0 release in 1-2 weeks with the tournament version (with support for selected variants), and later perhaps a 3.1 one with some minor user-facing features like additional DXP options on request.