Avoiding draws

[Ozymandias]

To get wins, we just need to raise the starting advantage after the opening moves. We've seen it at TCEC with the strongest engines and HW at play.

To bring those openings to correspondence and Freestyle, we just need players to have an incentive.

The idea would be to analyze all positions up to move 2 (less than 200K) and offer points for those positions above certain thresholds.

The organizers would be responsible for running the analysis and publishing the results for every player to place their bets/take their risks.

Two things would happen:

• many would move away from the most widely used openings
  some would overdo it and eventually lose games

[Modern Times]

Why do we need to avoid draws ? Is there something wrong with a draw ?

[chrisw]

Why do we need to avoid draws ? Is there something wrong with a draw ?

One reason be that current training algorithms require win/lose results in order to progress. If all games were draws it would not be possible to then make progress. Ozy’s idea attempts to get round this.

[Ozymandias]

Engine testing doesn't suffer from draws so much. Testers can define whatever imbalance they want and play games from there. The idea is about bringing THAT to centaur competitions. If nothing else, we'd get tie-breaks. But I'm sure we'd get much more.