Benjamin 2.0 - Tactical Analysis Engine

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Frank Quisinsky
Posts: 6808
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Frank Quisinsky »

Hi Ed,

that is a nice program for self playing!
A good idea to try out to added it in DGT Pi setups.

Again Ed do a great job for us.
Will be an event for chess computer freaks ... think so!

Thanks Ed!

Best
Frank

Small story:
For a while I had a dicussion in a private chat about such programs for self playing with ... (can't give the name). A weak later with a second GM about the same topic. Both asked for a strong aggressive chess programs with max. 2.600 Elo. Of course, Wasp on DGT Pi (strenght can be adjust from 1.500 - 2.700 Elo on DGT-Pi). Both Grandmaster's told me that Benjamin is nice but I don't know Benjamin. For self playing I am using M(T)D AnMon or Wasp. I found out that Benjamin comes from Ed Schröder and wrote ... yes, a program from Ed can't be bad. So Benjamin seems to be good known for grandmasters.

:-)
User avatar
Rebel
Posts: 6991
Joined: Thu Aug 18, 2011 12:04 pm

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Rebel »

Thanks Frank for your (always) kind words. Benjamin 2.0 has a 6th sense for tactics or finding unusual continuations, the cost for that is a serious loss in strength in normal games, 2 different diciplines. Maybe it's true what you say, that it might produce attractive self-play chess, I haven't tested that.
90% of coding is debugging, the other 10% is writing bugs.
Terje
Posts: 347
Joined: Tue Nov 19, 2019 4:34 am
Location: https://github.com/TerjeKir/weiss
Full name: Terje Kirstihagen

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Terje »

Rebel wrote: Wed Oct 21, 2020 8:31 pm
Terje wrote: Wed Oct 21, 2020 8:04 pm Did this improvement for analysis cost playing strength?
About 150-200 elo lower :lol:
Ouch, what did you change for it?
BrendanJNorman
Posts: 2526
Joined: Mon Feb 08, 2016 12:43 am
Full name: Brendan J Norman

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by BrendanJNorman »

Frank Quisinsky wrote: Wed Oct 21, 2020 11:11 pm

Small story:
For a while I had a dicussion in a private chat about such programs for self playing with ... (can't give the name). A weak later with a second GM about the same topic. Both asked for a strong aggressive chess programs with max. 2.600 Elo. Of course, Wasp on DGT Pi (strenght can be adjust from 1.500 - 2.700 Elo on DGT-Pi). Both Grandmaster's told me that Benjamin is nice but I don't know Benjamin. For self playing I am using M(T)D AnMon or Wasp. I found out that Benjamin comes from Ed Schröder and wrote ... yes, a program from Ed can't be bad. So Benjamin seems to be good known for grandmasters.

:-)
Very interesting. GMs are *finally* catching on to the idea of training against weak (but human-like) engines.

I love it.

For me, Benjamin is cool, but as far as Ed's work goes, my favs will always be ProDeo with ALEXAL, MACHEID or Storm personalities! :)

Storm is a killer!

EDIT: I also love ProDeo 2.6 and the ability to create your own personalities. I made a few of my own a couple years ago.
Frank Quisinsky
Posts: 6808
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Frank Quisinsky »

Hi Brendan,

GM knows the own weaknesses.
In most of cases the endgame.
The 16 World Champions we have ...
Most of the people are World Champion because they can play a better endgame as the others in the time they life.

So GMs are searching for a very strong opening phase, good middlegame and weaknesses in endgames. I say the human style, Wasp can play for an example (nice to see in FCP Tourney statistics by Klaus Wlotzka). GMs have more fun to play games vs. engines if they know that engines plays with full power. They like Stockfish as analysze engine but never for playing games itself vs. Stockfish.

Benjamin have a nice style of chess and one of the GMs with around 2.625 Elo have a lot of fun to play with it blitz games. He told me, that's a nice training, not more not less. So the gramdmaster is looking for others programs with the same playing style most of stronger humans have.

Spark ist here perfect with his 2.750 Elo and the gigantic king safty and very nice attacking style. But 2.750 Elo from the last Spark by Allard Siemelink is to strong!!

My problem all the time is that I can't know all the available chess programs.
So, Benjamin was new for me.

All the years computer chess I play my games vs. SSEChess or Phalanx and AnMon, vs. older Zarkov versions.
Since John added for Wasp 1.500 - 2.700 levels for DGT Pi, I am playing most of my own games vs. 2.150 Elo Wasp level and I am happy with it. Benjamin is since weaks in my self play directory too.

:-)

Best
Frank

I think BugChess have all that what Grandmasters search and is on the same level for the most of it with full power of strength. Thorsten Czub wrote about engines with a nice style and have BugChess on the own list. I wonder that Thorsten know that. Aristarch or ETChess too (two more examples).
BrendanJNorman
Posts: 2526
Joined: Mon Feb 08, 2016 12:43 am
Full name: Brendan J Norman

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by BrendanJNorman »

Frank Quisinsky wrote: Thu Oct 22, 2020 8:25 am Hi Brendan,

GM knows the own weaknesses.
In most of cases the endgame.
The 16 World Champions we have ...
Most of the people are World Champion because they can play a better endgame as the others in the time they life.

So GMs are searching for a very strong opening phase, good middlegame and weaknesses in endgames. I say the human style, Wasp can play for an example (nice to see in FCP Tourney statistics by Klaus Wlotzka). GMs have more fun to play games vs. engines if they know that engines plays with full power. They like Stockfish as analysze engine but never for playing games itself vs. Stockfish.

Benjamin have a nice style of chess and one of the GMs with around 2.625 Elo have a lot of fun to play with it blitz games. He told me, that's a nice training, not more not less. So the gramdmaster is looking for others programs with the same playing style most of stronger humans have.

Spark ist here perfect with his 2.750 Elo and the gigantic king safty and very nice attacking style. But 2.750 Elo from the last Spark by Allard Siemelink is to strong!!

My problem all the time is that I can't know all the available chess programs.
So, Benjamin was new for me.

All the years computer chess I play my games vs. SSEChess or Phalanx and AnMon, vs. older Zarkov versions.
Since John added for Wasp 1.500 - 2.700 levels for DGT Pi, I am playing most of my own games vs. 2.150 Elo Wasp level and I am happy with it. Benjamin is since weaks in my self play directory too.

:-)

Best
Frank

I think BugChess have all that what Grandmasters search and is on the same level for the most of it with full power of strength. Thorsten Czub wrote about engines with a nice style and have BugChess on the own list. I wonder that Thorsten know that. Aristarch or ETChess too (two more examples).
Hi Frank,

About the engines you mentioned, I certainly agree with everything you write. I am not too familiar with BugChes and will give it a try.

About the "weakness of GMs" being the endgame, I disagree.

You cannot generalize with human GMs as you might be able to with some computers.

Of the world champions, how can we possibly say that Capablanca, Smyslov, Fischer or Karpov's weakness was endgames?

And of non-World Champions, how can we say that Andersson, Miles, Korchnoi, Averbach, or Rubinstein failed to become World Champion because their endgame was too poor?

All of the above-mentioned players were endgame maestros and it was indeed one of the strongest parts of their game.

Here's how I see it:

At the 2500-2600 level they outplay opponents *mainly* based on superior positional judgment and more precise calculation.

At the 2700-2800 level, they outplay opponents also based on the above, but also opening preparation, extremely precise and high level of defense (Kramnik spoke of this "extraordinary defensive level" as the main change in opponents when he was first invited to attend Linares events in the early 1990s. "It is so hard to beat these guys").

At the 2800+ level, they can basically do everything and have no obvious weaknesses. For Caruana, his weakness is time management, which is more psychological than chess-related, and why his blitz ratings are much lower than standard. And ALSO why Carlsen deliberately went for the fast tie-breaks in their match.

Endgame skill is high for mostly all GMs, with some special guys (like those I noted) being standout geniuses.

Note: None of what I have noted above is comparing the players to engines, but to other humans.

For even a human GM to play training games against an engine, it must be no more than 2800 CCRL (2500-2700 is ideal IMO...so Phalanx 25 level to about Zarkov 6.55 level or a weakened Rodent) for the GM to even have a useful training session.

Engines are too strong now.
Frank Quisinsky
Posts: 6808
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Frank Quisinsky »

Hi Brendan,

I means that the advantage the World Champions have to other grandmasters are in my opinion more the strength in endgames. Good examples are the World Champions you numerate.

In my opinion strongest chess players in the World are in the first playing phase not a lot weaker as best chess programs. Most of strongest chess players know very exactly the opening theory and the ideas for the different opening systems. That give a small advantage in the beginning of the games. Often Carlsen comes with fantastic new ideas and very aggressive pawns. Carlsen give all that the cherry on top.

After different discussion with grandmasters they know that chess programs are in the very late middlegame and endgame not to beat. They speak itself from own weaknesses in comparing to the other playing phases. Most have to do with the pawn structure, build in the middlegames. A little secret is allways the pawn structure.

In the past we have a lot of good players, can be World Champion too.
Andersson (not Miles I think but one of my favorite players too), Kortchnoj.
Much more Ivantchuk or Speelman and so many others.

But Ivantchuk (my absolutely number 1) have to many bad times in his career. He plays often like a beginner and often like a Genius. Gelfand is such a phenomeon Kortschnoj was. Such a long time he plays chess on TOP niveau. I like his books "Technical Decision Making in Chess" and Dnymic Decision making in chess". At the moment I read the new book "Technical decision making in Chess". Really great material Gelfand give us.

But all in all ...
Not without a reason are players like Symslov, Fischer, Aljechin and I think 12 of 16 others World Champion.
They plays the endgame better as the others. That's the final result for myself after I study so many of chess books.

The other four are Tal, Kasparow and Carlsen.
Kasparow and Carlsen are pure allrounder's and after my opinions the strongest of all.
Tal with his great style in middlegames. I think for endgames much more players can play the endgame better as Tal in the time Tal life. But Tal comes with a fantastic style of chess.

Euwe was a very strong player and wrote fantastics chess books but from all the World Champions in my opinion the weakest.

Capablanca and Karpow the strongest World Champions in positional chess ...
Aljechin I like a lot, a very interesting person, such interesting for me like Karpow.

Best
Frank
Frank Quisinsky
Posts: 6808
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Frank Quisinsky »

Example to Ivanschuk ...
I try to find out that with ChessBase tactial analyzes.
Not the 100% best way ChessBase goes with that option (to many holes) but indeed interesting.

I made analyses from all the games of World Champions and other very strong players.

Now I build the TOP-20 with average of best 100 games with tactical analyses with Stockfish 10 and 11.
And on place 1 with a big advantage are the 100 best games Ivantschuk plays. Carlsen on two and Soe on three.

Interesting is that Capablanca is on 5 and Kasparow on 4.

Speaking from the tactical average from 100 best games.

Best
Frank

PS: Not a wonder that Carlsen have a lot of respect from Soe.
Frank Quisinsky
Posts: 6808
Joined: Wed Nov 18, 2009 7:16 pm
Location: Gutweiler, Germany
Full name: Frank Quisinsky

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Frank Quisinsky »

The tactical analyes from ChessBase is very interesting.

For an example:
Please use the fast won games from Fizbo with black (below move 50) and make the tactical analyses with Stockfish.
Stockfish found often not the right way ... Fizbo found!

Only one example!

Best
Frank
User avatar
Rebel
Posts: 6991
Joined: Thu Aug 18, 2011 12:04 pm

Re: Benjamin 2.0 - Tactical Analysis Engine

Post by Rebel »

BrendanJNorman wrote: Thu Oct 22, 2020 4:51 am Very interesting. GMs are *finally* catching on to the idea of training against weak (but human-like) engines.

I love it.

For me, Benjamin is cool, but as far as Ed's work goes, my favs will always be ProDeo with ALEXAL, MACHEID or Storm personalities! :)
The MACHEID (whatever that name means!) personality is also availabable as a second and next best personailty in Benjamin 2.0, I like it too, less agressive, but more reliable.
Storm is a killer!
Storm is ?, I don't have it, can you post?
EDIT: I also love ProDeo 2.6 and the ability to create your own personalities. I made a few of my own a couple years ago.
90% of coding is debugging, the other 10% is writing bugs.