Searching for more info on Monty

[GenoM]

Last version of Monty is an engine I recently found and immediately fell in love with it.
What an engine, what a style!
I'm very impressed. I think I've found Saint Graal of computer chess -- engine that plays so inventive that I'm astonished. To watch its games is pure pleasure!
Only thing that I know about it is that it uses MCTS.
Can someone share more about the engine and its authors?

[Frank Quisinsky]

Really a highlight in engine development!

A young programmer from England with ... in German we like to say ... Haare auf den Zähnen!
Give me contra in TalkChess for around 2 years in a thread. I am thinking OK, let see ...

As an old grandpa, I slapped me against the wall with a comment what are basics today.
OK, not very complicated to do that.

The future belongs to the younger generation. He is a prime example of this.
He is probably very talented and will fly the flag for computer chess with his work.
That's a good thing!

He provides the basics!
At the time, I was also very disrespectful. Today, that has completely changed.

[Ipmanchess]

https://github.com/official-monty/Monty
https://github.com/official-monty/Monty ... ase-latest

[Rebel]

Last version of Monty is an engine I recently found and immediately fell in love with it.
What an engine, what a style!
I'm very impressed. I think I've found Saint Graal of computer chess -- engine that plays so inventive that I'm astonished. To watch its games is pure pleasure!
Only thing that I know about it is that it uses MCTS.
Can someone share more about the engine and its authors?

I have tested Monty-20260111, it's strong and it is doing well at style testing.

Code: Select all

id name Monty-dev-20260111-53fb45f0
id author Jamie Whiting, Viren & The Monty Authors

Code: Select all

PGN   : pgn\uho-monty.pgn
Games : 15000
Won   : 9565
Time  : 5 seconds

 STYLE   King          Short           
 Total  Attack   %     Games   %     Sacs    %    Engine
249459   83456 19.5%  131336 63.8%   34667 14.1%  Monty 20260111  

Very nice score and I estimate an elo rating of +3450

I will add Monty to the playing style rating list, maybe it has a future over there.

I also tried Monty 1.0 (the original I think) but it too weak for the 3400 elo pool I use, it should get an absolute minimum of 100 won games out of the 15.000 games to get a style rating and it did not happen.

[Sylwy]

Well, a new logo for this superb chess engine ! Ruxy & ChatGPT, of course !
Like news: in my TOP_TEST_2026 I deregistered the engine called Revolver (0 points in 52 games for this "high-performance UCI-compliant chess engine") and replaced it with the latest Monty. For show........ !

[GenoM]

Firstly, thank you all for the replies.

Secondly, I want to share some thoughts about the way computer chess is moving forward.

I'm not a software engineer, just a fan of chess.

Intuitively I feel contemporary Stockfish went wrong way.
When it plays against the engines that are made in similar way (and of similar strenghth) it plays boring chess.
It is extremely STRONG and makes no mistakes, but chess it is playing is boring.

Imagine they finally reached their goal and made the Ultimate Stockfish version/chess engine that solves game of chess.
You start it and it shows: 1. e4 (+0.01, d1000), 1.d4 (-0.01, d1000) and so on.

What's the hell could be the use of that "knowledge"? It would be mostly useless, if you ask me. So I think it's a some kind of evolutionary dead end.

Some years ago I proposed here in the forum that a goal of the chess engine developers shouldn't be a making of the strongest chess engine but instead making an engine that would play like human -- and not in the sense of "errare humanum est" but about understanding chess ideas and plans and using logic that we can follow -- just like a strong human players do.

That my "proposition" wasn't met with enthusiasm and I let it go. A few years later, AlphaZero appeared and showed everyone that there is another way for computer chess to develop. But the chess engine developers took from AlphaZero only what was useful to them in pursuing their goal -- to make the strongest chess engine possible. I also think that copying Stockfish' way most of the chess developers took the wrong path.

But Monty is something VERY different. Monty's style (closest to AlphaZero's one to my eyes) is beautiful.
Sometimes it finds unexpected ideas and sometimes it plays really stunning moves.
Even its time management is weird in some human way: it thinks much over moves A/B engines would play in seconds and in every game it finally always fells into time trouble -- and then it plays very very good.
And something that is somehow irritating -- when it is winning its realisation is very weird too: I have the feeling that it delays the win purposefuly Like a human player that says: "My position is so winning that I can beat you even if I play not perfect moves" And finally it wins.

So I guess it's MCTS that brings this into the game so we need to follow the path taken by the authors of Monty.
I also think that if as much effort is put into the development of Monty by the computer chess community as was put into improving Stockfish, it will be extremely fruitful as a result.

[tapio]

@Archimedes, would an Android compile be doable?

[cpeters]

..nonsense.
sorry, wrong thread.