Question to Aart Bik

[abik]

Great to hear, that you already work on the FICS client!!!!!!! I will love the possibility to play with real human people on a wooden board. Thank you so much for your hard work!!! Thank you Aart!

Just keep an eye on this forum for progress updates on this.

I would prefer an opening book clubplayer that plays an opening suboptimal like a 2000 elo player. Perfect would be clubplayer-standard, clubplayer-passiv, clubplayer-activ, clubplayer-aggressive! I am just a fanatic chessplayer with a long wish list. I am not a software expert I dont know what is possible, how much work is it ...

I can't say I have much experience with book building, other than constructing a small built-in book for Chess for Android. Incidentally, this book consists of all the opening lines I had to study as a very young club player, so you may find the lines entertaining and not too overwhelming (enable the GUI book in the options menu, and disable the engine book in the engine set up window when playing each engine for the first time).

Below you will find estimated ELO ratings of the Android engines that participated in the 2013 tournament. Although these ELOs do not directly correspond to human ELOs and although this list is by now heavily outdated, it at least shows you should be able to find an engine with a comparable strength. I find it harder to say what engines have an interesting playing style. Unfortunately, I hate to admit, the bikjump engine is not one of them yet, it plays a rather crude game (something I still want to improve when all this GUI work finally settles down....). Perhaps others can chime in on interesting playing styles.

Code: Select all

1    Critter 1.6a 32-bit      2891
2    Stockfish 2.3.1          2854
3    Toga II 3.0              2763
4    Gaviota v0.86            2689
5    Komodo DEV               2670
6    Toga II 1.4.1SE          2650
7    DiscoCheck 4.0.1         2635
8    Arasan 15.2 JA           2601
9    MinkoChess arm           2593
10   Texel 1.01 32-bit        2575
11   crafty-235-android-ja    2568
12   Arasan 15.1              2564
13   Rotor 0.7                2549
14   Arasan 14.0.1            2535
15   Rodent 0.18.0            2527
16   cheng3 1.07 JA           2524
17   Daydreamer 1.75 JA       2505
18   scorpio-27-ja            2487
19   Pepito v1.59             2479
20   GarboChess 3             2467
21   Rotor 0.8                2465
22   sloppy-023-ja            2445
23   betsabe-II-130--ja       2429
24   danasah-488-ja           2422
25   phalanx-xxii-ja          2421
26   tucano104-ab             2406
27   typhoon-android-ja       2404
28   GNU Chess 6.0.2          2383
29   RedQueen 1.1.2           2367
30   greko98-ab               2363
31   Sungorus 1.4 JA          2350
32   TJchess 1.01U            2349
33   Jazz 6.40 JA             2344
34   natwarlal-014-ja         2316
35   scidlet-261b2-ja         2316
36   myrddin-086-ja           2316
37   kmtchess-121-ja          2305
38   gk-090-ja                2232
39   sjeng-112-ja             2224
40   BikJump v2.1P            2209
41   ZCT-0.3.2500             2184
42   AdroitChess 0.3          2136
43   tscp181-ab               2103
44   Diablo 0.5.1             2103
45   exchess-054-beta-ja      2101
46   zzzzzz-351-ja            2093
47   fairymax48l-ab           2076
48   surprise-43-b13-ja       2054
49   myrddin0.85j             2041
50   umax48w                  2019
51   VIRUTOR CHESS            1973
52   Chess for Android        1909
53   toledo-nanochess-2010-ja 1813
54   Simplex 0.9.8            1725
55   olithink-531-ja          1724

[PeterO]

Thank you Aart,

Is it possible to replace the gui book?

Peter

[abik]

Is it possible to replace the gui book?

Replace? Not at the moment, no. Disable? Yes, in the options menu, make sure the "GUI Book" option has no checkmark. In that case, engines will consult their own book (if any) right from the start.

The obvious next feature request is of course plugging in a new opening book in the GUI using one of the common opening book formats out there. This request is not very high on my TODO list, however.

Note that while playing engine tournaments (perhaps not of interest to you, but I am saying this just for completeness), there is an option to play all games from both sides of an external opening book. This allows you to create a tournament from, for example, all the Nunn positions or the Silversuite gambit opening suite.

[Ras]

Btw., what is the status with engine specific options? In CFA 2.5, they were not fully implemented, see http://www.talkchess.com/forum/viewtopi ... 81&t=37472 :

Even though the UCI standard states that the GUI should parse all options and build a dialog for the user to change settings, I will take a much easier approach in the next release of Chess for Android (...). I have picked a few common options (...) and present these to the user (...).

Does this still apply?

[abik]

Btw., what is the status with engine specific options? In CFA 2.5, they were not fully implemented, see http://www.talkchess.com/forum/viewtopi ... 81&t=37472 :

Even though the UCI standard states that the GUI should parse all options and build a dialog for the user to change settings, I will take a much easier approach in the next release of Chess for Android (...). I have picked a few common options (...) and present these to the user (...).

Does this still apply?

Yes, that has not changed much since then, although I probably added a few more built-in options since then. But the approach is still that the GUI assumes a few common options and provides widgets for these, and not the other way around as required by the standard (engine tells all options and GUI builds the widgets). I am interested in improving this, but probably also at a lower priority.

By the way, I *love* getting feature requests. But, conversely, I hope all Chess for Android fans understand I am doing this in my spare time next to a full time job and lovely family obligations.

So please be patient and understanding if I don't get to all requests

[Nordlandia]

I PM'ed Stefan-Meyer Kahlen (inventor of UCI protocol). He replied once he got time he will modify uci for hourglass format. That is good news

[abik]

I PM'ed Stefan-Meyer Kahlen (inventor of UCI protocol). He replied once he got time he will modify uci for hourglass format. That is good news

Great! So then we will have hourglass support for UCI and PGN. Did you initiate similar lobbying efforts in the XBoard/Winboard community to get hourglass time control added to the Chess Engine Communication Protocol (I don't think I have seen hourglass support in that standard yet)?

[Nordlandia]

Great! So then we will have hourglass support for UCI and PGN. Did you initiate similar lobbying efforts in the XBoard/Winboard community to get hourglass time control added to the Chess Engine Communication Protocol (I don't think I have seen hourglass support in that standard yet)?

I can initiate a thread about hourglass format on winboard / xboard forum. I think hourglass in combination with ponder and adjudication rules can as a whole be quite exciting and fresh.

[hgm]

If I understand correctly what hourglass TC means, it is a pretty funny TC. Once the opponent uses up almost all his time, you can basically dictate the speed of the rest of the game. Because when you keep moving very fast, he must follow that.

This suggests that it doesn't matter much how long you think. As long as you stay within time for the current move.

Does this really require a new TC mode, seen from the engine? It seems to me that the engine should pretty much do the same as it would with "movestogo 1 wtime T". There it knows that there is a hard limit for this move, and that it will derive some time advantage by not using up all the time. It doesn't seem very important to know if that time advantage will come from the remaining time being added to its own time for the next move, or subtracted from the opponent's time for his move.

If there is a difference, it is very subtle, and it is unlikely engines would want to take advantage of that to slightly increase their Elo at this TC (which almost no one would use). But if they would be so advanced that they have a way to exploit it, they can easily detect themselves that they are subject to this TC. Engines typically do measure how long they think, and they can compare this to the expected change in their clock time. The TC can also be represented (by convention) as "wtime T_clock winc T_hourglass/2", and if the engine detects a discrepancy in the way the wtime and btime change compared to what it knows it and its opponent have used, which approximately fits the hourglass model, it can invoke its special time management for this situation.

[abik]

Does this really require a new TC mode, seen from the engine? It seems to me that the engine should pretty much do the same as it would with "movestogo 1 wtime T". There it knows that there is a hard limit for this move, and that it will derive some time advantage by not using up all the time. It doesn't seem very important to know if that time advantage will come from the remaining time being added to its own time for the next move, or subtracted from the opponent's time for his move.

Thanks for your insights, H.G.

I think you are right that the GUI could simply implement it this way for xboard and also that most engines probably will just use up (part of) the remainder time. I just thought it would be a little cleaner to provide the engine with the exact time control details, just at the off chance some engine author has clever ideas how to exploit hour glass.....