nTCEC Season 2 information + changes

[Martin Thoresen]

*Only engines that are SMP capable with minimum 8 core support are participating.
*Engine updates are not allowed during a Season.
*The time control has been reduced from 150' + 60" to 120' + 30".
*6-men Nalimov has been removed since several engines have shown some strange behavior during the testing.
*The GUI will now adjudicate tablebase positions.
*The win/loss adjudication rule has been extended to 4 moves instead of 3. Still +/- 6.50 pawns.
*Stage 2a/2b has been merged into Stage 2, which is a single round robin.
*All openings will now be from recent human high rated tournaments to reflect the format of each Stage and there will not be any computer analysis of these positions, except to remove duplicates.
*The length of the openings have been reduced from 8 moves to 6 moves.

In addition, there have been some changes to the website etc:
*All games playing will be shown at http://tcec.chessdom.com/
*The chess board now has coordinates.
*The engines will now show distance to mate instead of very high evals.
*The chat has been moved to the top menu (will open in a new window).
*The widescreen option has been removed.

nTCEC Season 2 starts August 26th.

The list of participants is still not finalized, and will be published as soon as possible.

[Modern Times]

Fantastic Martin. Personally I'd have preferred a slightly shorter time control to make the games easier to follow. But I'm happy with your expert judgement, and I think 120' + 30" was also the general preference in your forum vote.

[Codesquid]

Sounds good, I'm looking forward to it. I hope that this time, Octochess makes to to stage 2

[Desperado]

Great ! i'm looking forward to this fantastic event.

regards
Michael

[Vinvin]

*Only engines that are SMP capable with minimum 8 core support are participating.
*Engine updates are not allowed during a Season.
*The time control has been reduced from 150' + 60" to 120' + 30".
*6-men Nalimov has been removed since several engines have shown some strange behavior during the testing.
*The GUI will now adjudicate tablebase positions.
*The win/loss adjudication rule has been extended to 4 moves instead of 3. Still +/- 6.50 pawns.
*Stage 2a/2b has been merged into Stage 2, which is a single round robin.
*All openings will now be from recent human high rated tournaments to reflect the format of each Stage and there will not be any computer analysis of these positions, except to remove duplicates.
*The length of the openings have been reduced from 8 moves to 6 moves.
....

Hello Martin !
IMHO, a lot of good changes !
Last year results here : http://chessengines2.blogspot.co.uk/201 ... s-for.html

Some suggestions/remarks :
- In stage 1, "9 rounds swiss" instead of "7" to have a better "middle of the pack" ordering (even the 2nd, Vitrivius was too high). And more than 32 engines is possible.
- "*The GUI will now adjudicate tablebase positions." -> this will be a bit random, I'd prefer let the engines play until the end (except clear opposite color bishops draw)

[Ajedrecista]

Hello Vincent:

Some suggestions/remarks :
- In stage 1, "9 rounds swiss" instead of "7" to have a better "middle of the pack" ordering (even the 2nd, Vitrivius was too high). And more than 32 engines is possible.

I am neither in favour nor against your suggestion. I only want to bring here a post from nTCEC forum:

http://www.tcec-chess.net/viewtopic.php ... rt=10#p607

The very minimum number of rounds to determine a unique winner through Swiss is the binary logarithm of the number of participants, which is 5 rounds for 32 engines. However, we need to find a set of qualifiers for the next stage, and for this, Square root of participants plus one is the formula that I know tournament directors to use, 7 rounds in this case for 32 engines.

The principle of Swiss is to pair evenly matched players every round, and after a certain number of rounds, it will not be possible any more & there will be pairings across different levels as the evenly matched players have already met - which I believe will destabilize the standings.

On a more technical/mathematical side, please take a look to the following post of the same thread:

http://www.tcec-chess.net/viewtopic.php ... rt=10#p693

Good luck to Season 2! Thank you very much for your effort, Martin.

Regards from Spain.

Ajedrecista.

[Martin Thoresen]

Hello Martin !
IMHO, a lot of good changes !
Last year results here : http://chessengines2.blogspot.co.uk/201 ... s-for.html

Some suggestions/remarks :
- In stage 1, "9 rounds swiss" instead of "7" to have a better "middle of the pack" ordering (even the 2nd, Vitrivius was too high). And more than 32 engines is possible.
- "*The GUI will now adjudicate tablebase positions." -> this will be a bit random, I'd prefer let the engines play until the end (except clear opposite color bishops draw)

Hi Vincent,

As Jesús pointed out, I am following the rule of thumb of the square root of the participants + 1 regarding Swiss tournaments. So that makes 7 correct.

The final number of participants remains to be decided on, but I can confirm that it will be more than 32.

Regarding the tablebase adjudication, yes I agree that the optimal is to let them play until the end, but again, I need to compromise - in this case in order to save time as the TC is still very long. So a long, shuffling endgame with 5 pieces or less will now be over immediately.

Best,
Martin

[Vinvin]

Hello Vincent:

Some suggestions/remarks :
- In stage 1, "9 rounds swiss" instead of "7" to have a better "middle of the pack" ordering (even the 2nd, Vitrivius was too high). And more than 32 engines is possible.

I am neither in favour nor against your suggestion. I only want to bring here a post from nTCEC forum:

http://www.tcec-chess.net/viewtopic.php ... rt=10#p607

The very minimum number of rounds to determine a unique winner through Swiss is the binary logarithm of the number of participants, which is 5 rounds for 32 engines. However, we need to find a set of qualifiers for the next stage, and for this, Square root of participants plus one is the formula that I know tournament directors to use, 7 rounds in this case for 32 engines.

The principle of Swiss is to pair evenly matched players every round, and after a certain number of rounds, it will not be possible any more & there will be pairings across different levels as the evenly matched players have already met - which I believe will destabilize the standings.

On a more technical/mathematical side, please take a look to the following post of the same thread:

http://www.tcec-chess.net/viewtopic.php ... rt=10#p693

Good luck to Season 2! Thank you very much for your effort, Martin.

Regards from Spain.

Ajedrecista.

Thanks for you interest, Jesus.
Posts you pointed out have not the same topic as mine. They want to have the strongest player at first place but as I wrote, I suggest more rounds to have a better order in the middle.