Initially I started this test in December 2005 to answer the
following question which despite many tests I had not seen conclusively
answered:
Are leading CM personalities statistically better than the
default settings?
After completing this test to my satisfaction with a number of personalities, it
seemed logical to use the benchmarks obtained to find out how good the latest
promising personality developed really is. So CGBT continues and will test
further selected personalities on an ongoing basis. Anyone who develops a
personality is welcome to contact me to see if it can be
tested.
Tourney rules:
Every tourney needs some basic rules and mine are as follows:
All personalities (incl Default) will play 1,000 blitz games
against exactly the same opponents
The opponents are fixed and will not be updated for
subsequent releases to ensure consistency over time
Each personality will play 40 games against 25 different non CM opponents (CM
personalities will NOT play each other)
Personalities to be tested will cover both CM9000 and
CM10th releases
The first 200 games will be used as a qualification to
determine which personalities play the full 1,000 games. A personality must
score > 100 to continue to complete the test
The tournament will be compatible with CEGT blitz testing
conditions (as a former CEGT tester, I like them!)
