Rowen wrote: ↑Tue Jan 19, 2021 1:20 pm
Hello Mr Forster
Thank you for your work on Sargon, I enjoy playing against it and it seems to play quite a solid game and from my perspective it makes a good training engine as it is approximately my strength. It is clearly an old program but that is why It is still playable from a strength perspective, so if it doesn't understand repetition etc, that's a function of its age. If you improve on it, great, but if you don't, thanks for your work in porting to UCI.
I also enjoy using your Tarrasch gui , it is a good minimalist program that I use solely to play against engines. I was wondering if it could be easier to change Engine options (parameters). I am aware that there are ways and workarounds,which I do implement, so it isn't that important, but I thought Id just ask.
Once again, Thank you for your work.
Thanks for your support. The engine options in Tarrasch are embarrassing. I hope to address it one day. My priority is to improve the code base in the hope of attracting other programmers to help. In the years I've been working on it C++ has changed dramatically. I have used C++11 features to improve some aspects of the code, but there's a lot more technical debt to be worked through before it would be comfortable for others to work with I think.
Sometimes (often) I think I should just let it go and only fix bugs. The danger with making improvements is that you introduce new bugs, I don't have the resources to re-test the whole thing before every release. Today for example I am scrambling to put out a maintenance release because it turns out I introduced a new (and very embarrassing - a whole minor feature doesn't work at all) bug with my last release in November. Because the program remains a niche one, even quite significant bugs like this can go unreported for two months.
I am surprised you use it to play against engines. That was my original application too, but I've hardly touched that for years and years. When I came to play my little match against Sargon 1978 I found it wasn't ideal any more - things have moved on. I missed being able to "pre-move" as I am used to on Lichess for example. At least I am pleased that a few years ago someone convinced me to allow the option of making a noise (even if it's only the system bell) when the engine moves. I quickly realised the importance of this in blitz in particular and turned it on for my personal use for the first time.
Chessqueen wrote: ↑Wed Jan 20, 2021 6:55 pm
My friend from Puerto Rico
rated 1620 Sanchez Martinez Xavier decided to play 10 game at 10'+5" his final score was 4 lost, 3 wins and 3 draws . Which is good news, but he noticed that Sargon is probably near 1700 to 1730 in the middlegame stronger than him in tactics, but very weak in the endgame without EGTB.
Yes Sargon 1978 is very weak in endgames because of its broad shallow search and its lack of chess knowledge. It doesn't realise pawns are potential new queens and doesn't bother advancing them in the endgame (except more or less by accident). Sometimes (more luck than anything) one of its pawns gets close enough to the end zone that a new queen might appear in the next handful of moves, only then does Sargon get the message.
In general there are (at least) two ways for weak players to beat Sargon;
1) Don't do anything, be patient, swap at every opportunity, wait for the endgame.
2) Build a massive attack, sacrificing if necessary, Sargon won't see deep enough to understand the point.
Of course 2) is a lot more fun than 1) and in my match I was yielding to the temptation to take that approach, when 1) is more cold-blooded and would have yielded better results. The fact that the opening book I was using led to exciting positions made 1) doubly hard to indulge.