Help with Komodo 1.0 personalities.

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Ovyron
Posts: 1971
Joined: Tue Jul 03, 2007 2:30 am

Re: Help with Komodo 1.0 personalities.

Post by Ovyron » Fri Jan 05, 2018 2:38 am

leavenfish wrote:In any case...to those who might go 'ga-ga' over 35. Re5....Man-up and admit, it is just a big stupid move! It's bad. Period. All too often what some refer to as 'style'...is just poor chess.
What if it isn't? What if you had 32men tablebases and Re5 is just as good as the rest of the moves, in that the theoretical result of the game doesn't change?

Style exists, because for 99.999andothernines% of positions, objectively, there's no such thing as bad moves, or good moves, unless they change the result of the game with perfect play, so a style can choose between the Ruy Lopez or the Italian, or between castling Queenside or Kingside.

One thing that is very apparent is that the top engines with the strongest elo have become more or less homogenous in their styles, in that, either they all agree a move is best, or after interactive analysis they can be made to agree on any move by refuting the others.

But this chosen move isn't best, and we know, because Komodo 13, Houdini 8, or Stockfish 10 are going to play different, stronger moves, in those positions, just like we're playing stronger moves than Komodo 9, Houdini 4 or Stockfish 6...

It's a dog chasing its tail, and it has no ending, until we reach the draw barrier with diminishing returns and chess engines hit an elo block (A0 didn't lose a single game, that's alarm bells for this), we just keep pursuing that way...

But I am willing to gamble, that I am able to produce games, where top engines of today destroy their previous versions, and other set of games where those previous versions destroy even older versions, and if I deleted the names of the players, you wouldn't be able to tell the difference, so what progress was actually made?

You can't tell the elo of the engine by how it plays, you need specialized software that tells you such and such is stronger than such and such and by how much.

But style? Style is very apparent and jumps out of the screen to smack you! You don't need to know at what strength those engines are playing, and their difference in strength is irrelevant.

What matters is the games! Again!

And in this thread, I guarantee you, that more people saw the games Brendan posted, played them out, and enjoyed them, than people actually going through the games that, say, Graham Banks continues to play over the years.

The saddest thing that has happened to computer chess is people running games on tournaments just to extract math from them, produce PGNs of endless series of games that nobody cares to replay and watch, turning chess, and using so many time and resources into producing text files with moves nobody reads and that are just used to calculate some arbitrary elo difference and perhaps add them to some book statistics for better move choices.

Engine style is the answer to this, let's use chess engines to produce chess games we can watch and enjoy. That's why I'm in this side of the issue, being more interested in what Personalities can be created that play chess moves never seen before that no other chess entity chooses, than what engines join the top 10 in elo and kick Chiron out of it...

Let's bring back chess to computer chess.

BrendanJNorman
Posts: 1262
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Help with Komodo 1.0 personalities.

Post by BrendanJNorman » Fri Jan 05, 2018 3:35 am

leavenfish wrote:In any case...to those who might go 'ga-ga' over 35. Re5....Man-up and admit, it is just a big stupid move! It's bad. Period. All too often what some refer to as 'style'...is just poor chess.
I'm sure you must be a lot of fun at parties, hey buddy? :D ;)

No in all seriousness, this goes back to the artist vs scientist debate.

Chess culture contains two types of nerds:

Computer nerds and chess nerds.

Computer nerds like leaving the latest Stockfish build to play 1000 game match, never look at a single game, but analyze some mathematical outcome afterwards.

They talk more about hardware optimizing than chess.

Guys like us are chess nerds.

We are the guys who at a chess tournament, finish our game and present it to our chess friends saying :"Check out the sacrifice I got away with, the guy had no idea and lost in a few moves" as we stand around a crowded board in the analysis room.

And when a "nerd" standing nearby says "Actually` it is just a big stupid move! It's bad. Period." as he thumbs his latest Stockfish build on his latest Android device, the crowd of CHESSPLAYERS says "Dude, go play with your phone over there" - as they point to a lonely corner of the tournament hall.

A lot of people tell others that Picasso's work is "Ugly, and not objectively strong (in terms of painting craft)" but they sell for millions.

And back to the strength issue, this Komodo Kinghunter is able to beat engines that beat and/or drew games with Kasparov, Kramnik, Leko and co.

As you guys endlessly chase the latest hardware and the latest "builds", we can be content with tweaking a ten year old engine.

As AO has a lot of "computer nerds" lamenting the " the coming end of computer chess", we respond "huh? I haven't had this much fun in years!"

I honestly think the "artist" side is more interesting, but each to his own.

I'm surely very biased in this respect.

As Uly said, at the top, engines have a homogenous style - it is very vanilla and all of the interesting ideas (from a human chess perspective) being pruned.

You can't tell who is who when watching the games.

With the most stylish engines, despite being weaker, you can see who is who just from watching moves.

Even sub-sections of attacking engines, I can recognize Thinker, Komodo KH, Zappa Mexico DA and my aggressive Rodent personalities just from watching the MANNER in which they attack.

This is really interesting.

it's no different to being a Topalov, Nakamura, Tal or Grischuk fan.

To tell their fans that their moves aren't "sound" will get you sent to the corner.

:lol:

P.S and there are a lot of us "style lovers" - my site already gets more traffic than 99% of computer chess blogs (or even this forum), and more subscriptions each day than forums like this get members...

...so I think we're onto something.

User avatar
Ovyron
Posts: 1971
Joined: Tue Jul 03, 2007 2:30 am

Re: Help with Komodo 1.0 personalities.

Post by Ovyron » Fri Jan 05, 2018 4:16 am

BrendanJNorman wrote:Computer nerds like leaving the latest Stockfish build to play 1000 game match, never look at a single game, but analyze some mathematical outcome afterwards.
Indeed, I have been too busy and haven't been able to train RomiChess against engines with stylized choices to see the outcome despite my computer sitting idle, doing nothing while I sleep. It would be trivial for me to play a hundred games daily of RomiChess vs something to see what happens, but I just don't have the time to watch all those games...

Because I refuse to produce games if I'm not going to watch them, whenever I do this, I'll check every single move of every single game because the games are the point.

It's incredible the different standards that people have for human chess players and engine chess players! If Tal plays a dubious move and wins the game they marvel. If some crazy engine personality plays a dubious move and only wins because the other missed some defense, they're accused of winning with poor chess.

But it goes beyond that, the people that hold the rating lists will not allow engines with learning, that would improve over time, because they claim it would mess their whole list, because it must have fixed participants, even though the original elo list, the FIDE ratings doesn't have a single fixed participant!

Chess rating lists are full of humans that may play terrible on sundays, or come drunk to a game and miss a mate in one.... humans learn, the rating is never static, people improve and get worse, and there's some prime that humans reach that is our peak and then we go downhill...

Yet none of that ruins rating lists of humans, but somehow an engine that becomes stronger over time would ruin an entire chess engine list?? Give me a break! :lol:

BrendanJNorman
Posts: 1262
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Help with Komodo 1.0 personalities.

Post by BrendanJNorman » Fri Jan 05, 2018 4:20 am

Ovyron wrote:
BrendanJNorman wrote:Computer nerds like leaving the latest Stockfish build to play 1000 game match, never look at a single game, but analyze some mathematical outcome afterwards.
Indeed, I have been too busy and haven't been able to train RomiChess against engines with stylized choices to see the outcome despite my computer sitting idle, doing nothing while I sleep. It would be trivial for me to play a hundred games daily of RomiChess vs something to see what happens, but I just don't have the time to watch all those games...

Because I refuse to produce games if I'm not going to watch them, whenever I do this, I'll check every single move of every single game because the games are the point.

It's incredible the different standards that people have for human chess players and engine chess players! If Tal plays a dubious move and wins the game they marvel. If some crazy engine personality plays a dubious move and only wins because the other missed some defense, they're accused of winning with poor chess.

But it goes beyond that, the people that hold the rating lists will not allow engines with learning, that would improve over time, because they claim it would mess their whole list, because it must have fixed participants, even though the original elo list, the FIDE ratings doesn't have a single fixed participant!

Chess rating lists are full of humans that may play terrible on sundays, or come drunk to a game and miss a mate in one.... humans learn, the rating is never static, people improve and get worse, and there's some prime that humans reach that is our peak and then we go downhill...

Yet none of that ruins rating lists of humans, but somehow an engine that becomes stronger over time would ruin an entire chess engine list?? Give me a break! :lol:
Shall we produce our own ratings list, including learning engines and personalities as part of the mix? :twisted:

User avatar
Ovyron
Posts: 1971
Joined: Tue Jul 03, 2007 2:30 am

Re: Help with Komodo 1.0 personalities.

Post by Ovyron » Fri Jan 05, 2018 4:41 am

BrendanJNorman wrote:Shall we produce our own ratings list, including learning engines and personalities as part of the mix? :twisted:
I'd much rather have a list that measured an engine's style instead of its strength... it could be instead of the elo rating, the... uly rating :lol:

But seriously, it could be trivial and fast, because an engine's style shows, we don't need 1000 games to measure it, a few wins against another engine can suffice, we extract the score from the best looking moves that it makes for attractive games.

The hardest part is deciding how to award the styles of the engines, we can become at the top with queen sacrifices down to pawn sacrifices, but also, ignoring opponent's sacrifices... There's just so much to measure but we could start by making a list of the things we liked of some mindboggling game like the one you posted, ranking them from the one we liked the most to the least, assigning scores to those patterns, and looking for more such patterns in those games.

Then we can become chess geeks ourselves and obsess over creating a personality that maximizes our scores, so, an engine that wins 3 games and draws 7 would stand worse than an engine that wins 4 and loses 6, if those 4 were won with style, while normal rating list would favor the former hugely.

We could finally be able to tell if the Anti-GM setting of ProDeo really does improve its style or if it's only useful against humans.

These are all abstract concepts that I haven't been able to put on paper, I just hope it's not something like "you know it when you see it", it would really suck if an engine's style was purely subjective and there was no objective way to measure it!

leavenfish
Posts: 229
Joined: Mon Sep 02, 2013 6:23 am

Re: Help with Komodo 1.0 personalities.

Post by leavenfish » Fri Jan 05, 2018 4:55 am

Ovyron wrote:
leavenfish wrote:In any case...to those who might go 'ga-ga' over 35. Re5....Man-up and admit, it is just a big stupid move! It's bad. Period. All too often what some refer to as 'style'...is just poor chess.
What if it isn't? What if you had 32men tablebases and Re5 is just as good as the rest of the moves, in that the theoretical result of the game doesn't change?
"What if it isn't?" ???? That's just more nonsense. "What if" all you want in one hand and pour a box of Lucky Charms in the other...and all you are going to get is Lucky Charms in one and a lot of nonsense in the other. Seriously, spend more time getting better at the game instead of pushing engines around and you might be able to judge a blunder from a decent move.


I'm serious - Argue all you want....it's a bad move, plain and simple...by any measure of what constitutes a 'good' or 'bad' move.

'Style' or 'personality'...how a Karpov might play a position as opposed to a Petrosian as opposed to an Alekhine...that's 'style or personality'. They can be very different...but a horrible move is a horrible move, regardless of who plays it.

Now...if you are one to throw a bucket of paint on a canvas (actually THROW it) and call it art...well, we are arguing on totally different planes and must leave it at that.

BrendanJNorman
Posts: 1262
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Help with Komodo 1.0 personalities.

Post by BrendanJNorman » Fri Jan 05, 2018 5:38 am

leavenfish wrote:Seriously, spend more time getting better at the game instead of pushing engines around and you might be able to judge a blunder from a decent move.
Ok.

Can I ask a couple of questions Brian?

1. How strong are you at chess? (Just curious given the "get better at chess" request from you which came across as a little condescending).

2. Why would you presume that Lev Aronian uses an engine that is "several years old" because it "suits his style"?

3. Why was Anand using Hiarcs in an age when Rybka, Stockfish and Komodo were available?

4. Why are these elite players not aware that the engines they choose, play "weak chess"?

Just curious Brian, no need to get all angry.

Damir
Posts: 1999
Joined: Mon Feb 11, 2008 2:53 pm

Re: Help with Komodo 1.0 personalities.

Post by Damir » Fri Jan 05, 2018 8:35 am

How can I see parameters in Komodo 1.0 ? It says they are empty, when I try and open the exe in Fritz GUI ? :( :(

BrendanJNorman
Posts: 1262
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Help with Komodo 1.0 personalities.

Post by BrendanJNorman » Fri Jan 05, 2018 8:38 am

Damir wrote:How can I see parameters in Komodo 1.0 ? It says they are empty, when I try and open the exe in Fritz GUI ? :( :(
Can you post a screenshot of what you're seeing?

There should be an option in UCI options to choose a path for a .per file (which is really just a text file renamed).

User avatar
Rebel
Posts: 4322
Joined: Thu Aug 18, 2011 10:04 am

Re: Help with Komodo 1.0 personalities.

Post by Rebel » Fri Jan 05, 2018 8:41 am

Ovyron wrote:We could finally be able to tell if the Anti-GM setting of ProDeo really does improve its style or if it's only useful against humans.
Just for the record the Anti-GM setting doesn't exist any longer. It has been replaced with something better -

[Confusion = 100]

Doing a mouse-over this option in the Personality Creator and it states -

Confusion is an improved version of the ANTI-ANAND algorithm we used playing Vishy Anand back in 1998. The algorithm tries to complicate the position for tactical shots and then profit from the created chaos by outsearchng the human opponent. The default value of [100] means the algorithm is inactive, other values than [100] will activate the slgorithm. Values that make sense are in the range of 25-50-75-125-150. A value of [500] will make wild sacrifices and are fun to watch but bad for the overall playing strength.

Post Reply