On-line engine blitz tourney August

[Henk]

Well, wrong knowledge is usually much more detrimental than no kowledge at all.

Fairy-Max has nothing but piece values, centralization (a sort of shared PST used by P, N, B and K, with a parabolic potential), 6th and 7th-rank Pawn bonus, a very kludgy Pawn structure (penalty for moving Pawns if there isn't one two squares left or right of it), a Pawn-push bonus that increases with game phase, penalty for moving the King and penalty for moving a Pawn straight in front of a back-rank King (the latter two switched off in the end-game).

That is all, and it doesn't do so bad.

My first version of RuyDos was comparable with Fairy-Max in strength, and it even had a similar style. Its evaluation was just material plus some penalty for isolated pawns. The moment that I added piece-square tables (copied from my previous program Ruy-López) it started to beat Fairy-Max pretty consistently, through better center control and better development in the opening.

RuyDos probably has a really good search as well.

[AlvaroBegue]

Well, wrong knowledge is usually much more detrimental than no kowledge at all.

Fairy-Max has nothing but piece values, centralization (a sort of shared PST used by P, N, B and K, with a parabolic potential), 6th and 7th-rank Pawn bonus, a very kludgy Pawn structure (penalty for moving Pawns if there isn't one two squares left or right of it), a Pawn-push bonus that increases with game phase, penalty for moving the King and penalty for moving a Pawn straight in front of a back-rank King (the latter two switched off in the end-game).

That is all, and it doesn't do so bad.

My first version of RuyDos was comparable with Fairy-Max in strength, and it even had a similar style. Its evaluation was just material plus some penalty for isolated pawns. The moment that I added piece-square tables (copied from my previous program Ruy-López) it started to beat Fairy-Max pretty consistently, through better center control and better development in the opening.

RuyDos probably has a really good search as well.

Now it does (or at least it's not half bad), but I am talking about an early version of the program. The search back then was correct but without many bells and whistles.

The point is that I remember distinctly that piece-square tables was the feature that made RuyDos become clearly stronger than Fairy-Max.

[Guenther]

First draft of the plain history is available!

Very interesting. Didn't remember RM had scored 50% in its second tournament, a feat it's yet to repeat.

Thanks! It will be ready in a few days, because I am on a little trip tomorrow.

Guenther

The main page with the history of all result tables is ready now.
On top of the page there is an 'overview' which is linked by number
to each tourney.
(If one clicks on the header of a tourney he will be led back to the overview for conveniance.)

Stats probably at the weekend.

Guenther

[Guenther]

The main page with the history of all result tables is ready now.
On top of the page there is an 'overview' which is linked by number
to each tourney.
(If one clicks on the header of a tourney he will be led back to the overview for conveniance.)

Stats probably at the weekend.

Guenther

Half of the work on an 'eternal table' with stats and diagrams is finished.
http://rwbc-chess.de/HGM_tourneys.htm

I have a few questions concerning names of participants @HGM mostly:

Two times instead of 'Goldbar' a nick called 'Goldbaar' appears, is this
practically the same player or a second different account (or even a typo)?

Capivara started in its first tourney as 'CapivaraLK' but already in the
next tourney and all later ones there is a plain 'Capivara?

So far I have found also one account named 'admin'
with only one participation so far (after 53/100 tourneys).
Should I treat it different or combine with one of your entries?

For all other variations it is clear that it was done by users, almost
always Graham and Ray and I kept them divided (different hardware/books etc.).

Guenther

[hgm]

Cool!

For CapivaraLK the LK suffix means it was run by a user. Because the machine on which the ICS was running at the time suffered a head crash of its HD, and I had never made any backups, it isn't possible to check what 'LK' stood for from the finger notes anymore.

As for the participation of 'admin', this baffles me. 'Admin' is the account with ADMIN_GOD permission that I use to watch the tourneys and control mamer. The only reason I can imagine for having it participate in an engine tourney is that I quickly needed a replacement participant, and did not have any time to create an account for it. I don't recall anything of this, however. I certainly did not participate in person.

[Guenther]

Cool! 8-)

For CapivaraLK the LK suffix means it was run by a user. Because the machine on which the ICS was running at the time suffered a head crash of its HD, and I had never made any backups, it isn't possible to check what 'LK' stood for from the finger notes anymore.

As for the participation of 'admin', this baffles me. 'Admin' is the account with ADMIN_GOD permission that I use to watch the tourneys and control mamer. The only reason I can imagine for having it participate in an engine tourney is that I quickly needed a replacement participant, and did not have any time to create an account for it. I don't recall anything of this, however. I certainly did not participate in person.

I hoped you like it ;-)

If all is finished I have to judge which view would be best in 'read only' mode for the stats, because it looks better with some columns hidden (main table is huge in width).
I also need to check again what permissions are there in read only mode, may be I even allow full permission and backup a master hoping for no abuse in the tables.

May be I find more infos in the thread about the tourney with 'admin'.
(Normally you gave a little report already for unusual 'events' in the post with the result table, at least after the very first tourneys - those reports are added already)

A slightly related question BTW:

Code: Select all

/icsNames={80.100.28.169 /icshelper=timeseal
chessclub.com /icsport=5000 /icshelper=timestamp
freechess.org /icsport=5000 /icshelper=timeseal
winboard.nl /icsport=5000 /icshelper=timeseal
global.chessparlor.com /icsport=6000 /icshelper=timeseal
chessanytime.com /icsport=5000
chess.net /icsport=5000
zics.org /icsport=5000
jogo.cex.org.br /icsport=5000
ajedrez.cec.uchile.cl /icsport=5000
fly.cc.fer.hr /icsport=7890
freechess.nl /icsport=5000 /icshelper=timeseal
jeu.echecs.com /icsport=5000
chess.unix-ag.uni-kl.de /icsport=5000 /icshelper=timeseal
chess.mds.mdh.se /icsport=5000

I guess I have all those ICS adresses copied since long time with my ini file.
Do you have an idea which are definitely obsolete and if some are missing here?

Guenther

[Guenther]

May be I find more infos in the thread about the tourney with 'admin'.
(Normally you gave a little report already for unusual 'events' in the post with the result table, at least after the very first tourneys - those reports are added already)

found it:

Code: Select all

The stress test worked well; at one time we had 20 guests each observing all games, and no problems were detected. Zappa had to be forfeited because it could not stay connected longer than 1 min; this seems tohave beenalocalproblem, though, as it apparently had gone away when that machine was switched to another wireless adapter, and started observing games. The game Crafty-Gaviota did not start, which wasnoticedonly after the round completed (so we all had to wait for it...) but later inspection of the channel-49 traffic revealed Gaviota had been disconnected when the round started.

The mamer patch seems to work: I see no more 3 consecutive same colors or an imbalance > 2 anywhere in the schedule.

Oh, and 'admin' was Joker for the first 2 rounds, and replaced by Ayito for the rest (both on a slow laptop), when I discovered that observing a game made Joker crash.

[Guenther]

It is finished now and there are some diagrams too.

Here as little stats example the table sorted by participations/success
Of course HGMs programs with the most tourneys, but interesting
who follows after them.
BTW NightmareX together with Nightmare (which it superseded obviously)
would be even more constant than Goldbar.
Also Texel is remarkable despite its relatively young 'age' here.

Code: Select all

#       Name            Points  Games   RemGam  Tourn.  % tot.  T. Wins
1       microMax        198.5   817     0       93      24.3    0
2       Spartacus       286.0   689     0       78      41.5    0
3       Joker           291.5   688     0       77      42.4    0
4       Goldbar         392.5   671     0       76      58.5    1
5       WaDuuttie       352.5   558     0       63      63.2    0
6       Baron           265.0   467     1       53      56.7    0
7       ArasanX         283.0   448     1       50      63.2    1
8       Texel           319.0   434     0       49      73.5   15
9       Almere          253.0   429     0       48      59.0    2
10      Nightmare       284.0   408     6       47      69.6    9
11      Fizbo           257.0   360     8       41      71.4    5
12      NightmareX      215.5   332     0       37      64.9    2
13      Tinker          167.5   330     0       37      50.8    0
14      Skipper          52.5   314     0       35      16.7    0
15      Rookie          165.0   293     0       33      56.3    0
16      Myrddin         127.0   286     0       32      44.4    0
17      NebiyuHG        112.0   288     0       32      38.9    0
18      Zurichess       131.0   267     2       30      49.1    0
19      Tornado         146.5   259     0       29      56.6    0
20      Arminius        136.5   258     0       29      52.9    0
21      GaviotaRB       146.5   243     0       27      60.3    4
22      Thinker**       192.0   216     4       25      88.9   17
23      Gaviota         137.5   226     0       25      60.8    1
24      EXchess         130.5   215     0       24      60.7    0
25      Schooner        100.5   206     0       23      48.8    0

A small caveat: Not completely cross-checked yet for data inserting errors...

BTW when is the September tourney? ;-)