GM Vs Komodo Time Odds Match

[ChessRenewal]

I would like to see another match along these lines (perhaps giving Komodo less than three minutes):

https://www.chess.com/news/view/time-od ... 5-0-5-6926

In total, Erenburg received six different odds. These were as follows.

Erenburg was granted 30-1 time odds. He had 90 minutes plus a 30-second increment;

Komodo had three minutes plus a one-second increment.
Komodo was limited to a single core. Komodo normally uses 24 cores.
Komodo was not allowed to think on Erenburg's time.
Komodo's opening book was limited to three moves.
Erenburg had the white pieces in all games.
Komodo did not have access to tablebases.

[lkaufman]

I would like to see another match along these lines (perhaps giving Komodo less than three minutes):

https://www.chess.com/news/view/time-od ... 5-0-5-6926

In total, Erenburg received six different odds. These were as follows.

Erenburg was granted 30-1 time odds. He had 90 minutes plus a 30-second increment;

Komodo had three minutes plus a one-second increment.
Komodo was limited to a single core. Komodo normally uses 24 cores.
Komodo was not allowed to think on Erenburg's time.
Komodo's opening book was limited to three moves.
Erenburg had the white pieces in all games.
Komodo did not have access to tablebases.

I think that if we play another such match with a GM from the normal starting position, we would add the condition that draws count as wins for White, the GM. But then the problem is that if we play without a book we might get repeat games and/or games that follow theory for too long, whereas if we make a special three move book the results might depend more on that book than on the engine's play. I think that we could even add the handicap that Black forfeits the right to castle, which makes all book knowledge moot. Or we could require that Black's first move be 1..f6, probably the worst move that Black can commit to without blundering a pawn. In all cases White would be playing normal chess, only Black would have to do something weird. Of course with large enough time odds normal chess can be fair with anyone, that's basically what our levels 1 thru 24 are for, but it's not particularly interesting to learn whether a particular opponent needs 200 to 1 or 500 to 1 time odds (for example), which in turn depends on the hardware speed and how many threads are allowed.

[mwyoung]

I would like to see another match along these lines (perhaps giving Komodo less than three minutes):

https://www.chess.com/news/view/time-od ... 5-0-5-6926

In total, Erenburg received six different odds. These were as follows.

Erenburg was granted 30-1 time odds. He had 90 minutes plus a 30-second increment;

Komodo had three minutes plus a one-second increment.
Komodo was limited to a single core. Komodo normally uses 24 cores.
Komodo was not allowed to think on Erenburg's time.
Komodo's opening book was limited to three moves.
Erenburg had the white pieces in all games.
Komodo did not have access to tablebases.

I think that if we play another such match with a GM from the normal starting position, we would add the condition that draws count as wins for White, the GM. But then the problem is that if we play without a book we might get repeat games and/or games that follow theory for too long, whereas if we make a special three move book the results might depend more on that book than on the engine's play. I think that we could even add the handicap that Black forfeits the right to castle, which makes all book knowledge moot. Or we could require that Black's first move be 1..f6, probably the worst move that Black can commit to without blundering a pawn. In all cases White would be playing normal chess, only Black would have to do something weird. Of course with large enough time odds normal chess can be fair with anyone, that's basically what our levels 1 thru 24 are for, but it's not particularly interesting to learn whether a particular opponent needs 200 to 1 or 500 to 1 time odds (for example), which in turn depends on the hardware speed and how many threads are allowed.

I like time odds matches much better. I would like to see how much of a time advantage a GM would need to win a time odds match. I bet on one core, Komodo could win this match with 1 minute vs 90m+30s.