[Andy_L]

When a Robot Tournament does not have a full field consider letting "GIB" compete.

Add a GIB player to the list of human players so we can see GIB's result vs. ours. Of course, GIB would be a non-BBO masterpointing participant. It would be curious to me to see the GIB result posted... against the field of us humans. If the field were of a small enough size... add multiple GIB players to the field -say maximum of three (3). It would be curious to me to see the results of one GIB player vs. another GIB player; would they have identical results? How varied would their results be?

As a last benefit at least I would know the competing GIB would be subject same GIB play as humans are... and I'll be able to imagine in binary code somewhere one GIB cussing out another for a bad result on a hand.

[barmar]

Unless we force them to use different random number sequences, all robot-only tables would be expected to have the same results.

[cloa513]

barmar, on 2012-March-13, 15:09, said:

Unless we force them to use different random number sequences, all robot-only tables would be expected to have the same results.

But won't.

[Vampyr]

barmar, on 2012-March-13, 15:09, said:

Unless we force them to use different random number sequences, all robot-only tables would be expected to have the same results.

But there would only be one robot-only table.

[barmar]

Vampyr, on 2012-March-15, 21:05, said:

But there would only be one robot-only table.

Andy suggested "maximum of three". I thought the point of his suggestion is to add enough robot tables so that we would get reasonable matchpointing -- matchpointing in a small field is pretty random, because there's not much field protection (the smaller the field, the closer it is to BAM, generally considered the toughest form of scoring).

[Vampyr]

barmar, on 2012-March-15, 23:50, said:

Andy suggested "maximum of three". I thought the point of his suggestion is to add enough robot tables so that we would get reasonable matchpointing -- matchpointing in a small field is pretty random, because there's not much field protection (the smaller the field, the closer it is to BAM, generally considered the toughest form of scoring).

Oh I see. I missed the maximum-of-three part.

I think an excellent idea would be to use one or more robots, and give them matchpoints but not give matchpoints to players against the robots. It would be a way to find out how good the robots really are.