CDrill 1800

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
PeterO
Posts: 121
Joined: Sun Jul 31, 2016 4:35 pm

Re: CDrill 1800

Post by PeterO » Thu Apr 12, 2018 4:08 pm

Next idea,

If 2 possible moves are quite equal - decide by random!
(+/- 0,2. - or something like that).

Peter

GregNeto
Posts: 28
Joined: Thu Sep 08, 2016 6:07 am

Re: CDrill 1800

Post by GregNeto » Thu Apr 12, 2018 7:23 pm

I really like Cdrill 1800 very much because it is bug free (which is unusual for an engine in this elo region) and very well balanced. We tried it in our club and it plays a strong european 1800 when humans are using not too much time (blitz). It does not get stronger at longer time controls so humans getting better results when thinking longer. Comparing Human and Computer Elo is very hard if not impossible as we all know.

Ferdy, would it be possible to do a Cdrill 2000? Maybe just a simple search extension will do.

Frank Quisinsky
Posts: 4852
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by Frank Quisinsky » Thu Apr 12, 2018 8:07 pm

Hi Ferdinand,

I like it.

In my opinion important:

1. Mate extension (the way to the success Johan has with TheKing in chess computer times).

2. The opening book should be very strong. You can used the FEOBOS database (error free) as engine book if you like. So CDrill can play all 500 ECO codes.

3. Most chess players have a rating between 1.400 - 2.100 Elo.

So it would be great to have 8 levels.

Lev 1: 1.400 Elo
Lev 2: 1.500 Elo
Lev 3: 1.600 Elo
Lev 4: 1.700 Elo
Lev 5: 1.800 Elo
Lev 6: 1.900 Elo
Lev 7: 2.000 Elo
Lev 8: 2.100 Elo

I am thinking 1.800 isn't enough.
2.100 would be better with levels.

4. Chess program should have a mega aggressive style. So humans can try to win with her own positional style.

Loud thinking!

Best
Frank

CDrill could be interesting for TMET (Mephisto chess computer tourney I will start June 2018). So if you like to see it vs. all the older Mephisto chess computers ... please let me know. Maybe we can join two different ideas to a new one. So my tourney can be more interesting and the development phase for CDrill will get a small booster.

I don't know (idea is good or not) but if you are active I have each time interest on it after all your helps for myself and my own things. To give my support will be an event for myself.
I like computer chess!

User avatar
PeterO
Posts: 121
Joined: Sun Jul 31, 2016 4:35 pm

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by PeterO » Thu Apr 12, 2018 8:22 pm

I absolutely agrre with you Frank,

But NO levels. Lets set the elo as UCI option. I want to set the Elo direct - dont want to remember „Which level is Elo 1900?“

Peter

Frank Quisinsky
Posts: 4852
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by Frank Quisinsky » Thu Apr 12, 2018 8:45 pm

Hi Peter,

Elo is fine for me and I am sure for the most of others.


For Ferdinand:
But 1.800 Elo isn't strong enough. Program should have 2.100 Elo if ... Shredder 12 have 2.800 Elo (to compare better with the most of rating systems).

For examples:
Ratings for the older chess computers are to high (around 150-200 for SSDF). OK, not easy to build. Most problems here the endgames from most of the older chess computers. Impossible to create a rating if ...

Endgame for MMV = 1.400 Elo
late middlegame for MMV = 1.800 Elo
middlebame for MMV = 2.000 Elo
after openings = 2.100 Elo

The differents are to big in my opinion.

That is most complicated for Ferdinand with CDrill I think.

Humans have today no chance vs. chess programs in late middlegame / endgames. In chess computer times ... the chess computers have in most of cases no chance vs. humans.

A human like style produced Wasp!
Stronger in middlegames, weaknesses in transposition into endgame (in comparing to the other top programs). But on highest GM level niveau (Wasp).

CDrill should be have for a human like style:

- after openings (stronger)
- middlegame (stronger)
- transposition into endgame (weaker as in the phases before)
- endgame (a bit weaker as in the phases before but clearly better as the older chess computers)

Humans have to fight in opening phase and early middlegame. And the perfect chess program for humans was born ... with all this what humans like ... aggressive style, mate extensions.

Chess computer's by Richard Lang are in the near but the style is not aggressive enough. The only perfect chess computer in producing such an interesting style are in my opinion the latest programs by Dave Kittinger (Novag Diamond, Diamond II, Start Diamond, Star Diamond II).

Best
Frank
I like computer chess!

User avatar
Rebel
Posts: 4788
Joined: Thu Aug 18, 2011 10:04 am

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by Rebel » Fri Apr 13, 2018 6:48 am

A while ago I released ProDeo 2.4 but received hardly any feedback if there is value in the chosen framework for improvements.

http://rebel13.nl/prodeo/prodeo24.html

BrendanJNorman
Posts: 1486
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by BrendanJNorman » Fri Apr 13, 2018 7:00 am

Frank Quisinsky wrote:Hi Ferdinand,

I like it.

In my opinion important:

1. Mate extension (the way to the success Johan has with TheKing in chess computer times).

2. The opening book should be very strong. You can used the FEOBOS database (error free) as engine book if you like. So CDrill can play all 500 ECO codes.

3. Most chess players have a rating between 1.400 - 2.100 Elo.

So it would be great to have 8 levels.

Lev 1: 1.400 Elo
Lev 2: 1.500 Elo
Lev 3: 1.600 Elo
Lev 4: 1.700 Elo
Lev 5: 1.800 Elo
Lev 6: 1.900 Elo
Lev 7: 2.000 Elo
Lev 8: 2.100 Elo

I am thinking 1.800 isn't enough.
2.100 would be better with levels.

4. Chess program should have a mega aggressive style. So humans can try to win with her own positional style.

Loud thinking!

Best
Frank

CDrill could be interesting for TMET (Mephisto chess computer tourney I will start June 2018). So if you like to see it vs. all the older Mephisto chess computers ... please let me know. Maybe we can join two different ideas to a new one. So my tourney can be more interesting and the development phase for CDrill will get a small booster.

I don't know (idea is good or not) but if you are active I have each time interest on it after all your helps for myself and my own things. To give my support will be an event for myself.
Hi Frank, I agree with everything here except that the opening book should be very strong.

If the program is indeed designed for play against humans, it need not be strong at all, since most humans below 2100 are far from walking opening encyclopedias themselves.

In my opinion, the ideal situation is for it to have a few "repertoires" which can be selected in UCI options and these dictate how the engine will select openings.

So, if you choose in UCI options "positional repertoire", it
selected 1...e5, French, Caro Kann etc and other "solid" lines for black and maybe 1.d4 and 1.c4 as white.

An "aggressive" repertoire could come from 1.e4 as white (with things like 6.Bg5 or 6.Be3 vs Najdorf etc) and Sicilian, Scandinavian and so on, as black.

The opening repertoires can be wide, but should only be about 8-10 moves deep.

This will simulate the opening knowledge of the average 1800-2100 player quite well.

BrendanJNorman
Posts: 1486
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by BrendanJNorman » Fri Apr 13, 2018 7:06 am

Rebel wrote:A while ago I released ProDeo 2.4 but received hardly any feedback if there is value in the chosen framework for improvements.

http://rebel13.nl/prodeo/prodeo24.html
Hi Ed,

Can this Elo adjustable version use personalities from ProDeo 2.6?

The ones created with your online tuner, I mean...

I wouldn't mind trying some of the personalities I've created at say 2200 Elo to see the resulting style. 8-)

Frank Quisinsky
Posts: 4852
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: CDrill 1800 for TMET ... vs. the Mephisto's?

Post by Frank Quisinsky » Fri Apr 13, 2018 7:12 am

Hi Ed,

yesterday I am on your site for links I am searching to your work for Mephisto (found it). For my personal chess award.

Have a look on:
http://www.amateurschach.de/main/_chess-computer.htm

At the moment the older chess computers are working on a small test (8 chess problem, tuning test).

To your answere:
New for me, sleeping.

The problem with me is Arena.
I am not start the GUI since I gave up my work with Martin. I am working many years with much blood, sweat and tears on it. So it's better for me not to start the GUI since I gave up the work ... not important. I believe not easy to understand for others.

But I am looking what you do on your site.
Looks greats for me, idea for ProDeo 2.4 is fine.

PS: I have nothing against MMV. I like MMV a lot ... in my posting I wrote only that is not easy to create ratings for most of the older chess computers if a part of the game (endgame) much weaker than the others parts.

These days I have again a lot of fun with the work you do for Hegener & Glaser. Can't give my compliments often enough.

Mega IV and Super Mondial II is new for me.
In the past I never have it.
At the moment I am looking in my small test with Super Mondial II.

:-)

Best and have a good time!
Frank
I like computer chess!

Frank Quisinsky
Posts: 4852
Joined: Wed Nov 18, 2009 6:16 pm
Location: Trier, Germany
Contact:

Re: BTW: Short question to the black Mirage board!

Post by Frank Quisinsky » Fri Apr 13, 2018 7:33 am

Hi Ed,

BTW: A short question
Sorry to the others for my BTW question!

Mephisto Mirage board is working fine with MMI, Mirage and Rebell 5.0 modules (means not with the keyboard of modules, means with the sensors on the board).

Is this random or do you support the black Mirage board with an extra work?

Thats great, I like the black mirage board a lot.

I have two Rebell 5.0 modules.
One is tuned to 9.4Mhz. The tuned module is not working with sensors.

If you have time for an answere ... only for my personal interest on Mirage board.

...

:-)

Best
Frank
I like computer chess!

Post Reply