BT2450, BT2630

David Coons

unread,

Dec 14, 1996, 3:00:00 AM12/14/96

to

Two of the game databases that came with Rebel 8 were BT2450.DAT and
BT2630.DAT. The instructions book calls them BT tests but doesn't say
anything else about them other than that BT2630 is the "latest
BT-test". Can anyone explain how to use them?

Chris Whittington

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

They should contain 30 positions each.

Enter each position into your computer, and let it think for 15
minutes (900 seconds).

If a position is solved, write down its solution time in seconds.

It doesn't count as a solution if the program finds, the move
and then changes its mind.

If after finding a move, then changing its mind, then finding
it again, you should use the *last* time found.

Any solution that is not found, score as 900 seconds.

Add up all the times.

Divide by 30.

Subtract the result from either 2630 or 2450.

The final figure is the ELO rating.

Or, isn't the ELO rating,as the case may be.

Old tests get frigged by some programmers (they recognise the
position and apply a special kludge factor to 'help' the solution)

Old test are often used for tuning purposes. This results in programs
performing better, either by design or as a side effect.

30 positions are not enough.

Test positions favour 'finders' rather than 'planners'. Its quite
easy to make a program be a finder, very good at finding
combinations *if they are there*, but then it becomes hopeless
at chess games with many positions where there is nothing
to be found, just good positional moves.

Like all numbers trying to encapsulate complex behaviour BT test
results have their place. But ......

Chris Whittington

mclane

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

d.c...@ttu.edu (David Coons) wrote:

>Two of the game databases that came with Rebel 8 were BT2450.DAT and
>BT2630.DAT. The instructions book calls them BT tests but doesn't say
>anything else about them other than that BT2630 is the "latest
>BT-test". Can anyone explain how to use them?

We have 2 testers in germany that published nice test-suites in the
big-german COmputerchess-magazine Computer-schach and spiele.

One is named Hubert Bednorz, the other Fred Toennissen.
Bednorz is the B in BT. Toennissen is the T in BT.
The BT2630 is the predecessor of the BT2450-test-suite.

If you find all positions in the BT2630 in 0" seconds, you have 2630
ELO.

Same with BT2450-test-suite.

The BT2450 was too easy for the strong tactic programs, so B and T
exchanged some positions and the result is the BT2630 test-suite.

In my opinion the test works nice, but it only finds out about the
tactical strength of a program. I know many programs that do not have
a high BT-rate, but play strong chess.

V999

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

i can give an example of odd bt results. not long ago i posted some odd
bt results on some old programs. these games preceded the tests so they
could not have been rigged. on a p100, cm2000 scored 1823 and sargon4
scored 2024. i believe cm2000 is much stronger. as a quick check, i
played a blitz game and it wasnt even close.
lately there have been discussions about a chess aptitude test.
why not have a test with a mix of positions typical of a full game. it
could have realistic time limits which may vary according to each
situation (rather than 15 min for all like bt). it could give partial
credit for good moves. and it could give negative points for a overall
low score in a subset of positions that indicates a gap in a crucial area
(like endgame).
any comments?

thanks,
vince