[Bug-gnubg] Bug: 1-ply outperforming 2-ply

[Eric Duran]

Windows 7 64-bit pro
Version 1.05.000-mingw 32-Bit 20150725
AMD FX 8350 4GHz

I've run several CPU-vs-CPU matches with the following settings, and
the 1-ply player almost always outperforms the 2-ply player:

Player 1:
Checker Play
Lookahead: 1-ply
Cubeful evaluation
Neural-net pruning
Noise: 0
No move filter

Cube Play
Lookahead: 1-ply
Cubeful evaluation
Neural-net pruning
Noise: 0

Player 2:
Checker Play
Lookahead: 2-ply
Cubeful evaluation
Neural-net pruning
Noise: 0
No move filter

Cube Play
Lookahead: 2-ply
Cubeful evaluation
Neural-net pruning
Noise: 0

I've tried swapping the players, using different dice generators
(random.org, MT), but the 1-ply player almost ALWAYS beats the 2-ply
player on my machine, by A LOT, even winning by a gammon a fair amount
of the time. Very strange. Shouldn't the 2-ply player be a lot better
than the 1-ply player? I've tried running with & without NN pruning,
move filters, etc, but I always get the same result! What's going on?

Here's a typical 10-point match game where the 1-ply player beats the
2-ply player 11 to 6:

https://hexandoct.000webhostapp.com/1-ply-2-ply_10p_2017-03-12.sgf