Current status of the statistics for 6 pc EGTB's

[vb4]

How do we stand as far as the correct statistics in a format similar to Eugenes .tbs files? Have the issues been resolved as far as what is causing the decrepencies. Be interested in hearing any feedback here in this forum.

Thanks,

Les

[guido]

How do we stand as far as the correct statistics in a format similar to Eugenes .tbs files? Have the issues been resolved as far as what is causing the decrepencies. Be interested in hearing any feedback here in this forum.

Thanks,

Les

Hi Les,

I waited for the answers of someone else because I can speak only of the TBs generated by my program.
In these last times I generated some more 5-1 TBs but I'm still far from the end of the job (47 of 70) because I encountered some very long TBs in this group.
From the comparison with the results of Martin (FEG) discussed in another thread I never found differences which demonstrated an evident errors of one or both the generators, but only differences due to the criteria chosen for the legality of the positions or to the diagonal symmetry (both the kings on a1-h8 ).
Generally I eliminated more illegal positions than FEG and TBGEN, so the numbers in my statistics should be always less than the correspondent values of the other authors. If this doesn't happen, an error surely would exist but until now this has not happened.
But the differences in some specific cases are very big: for example in the KQQKQQ endgame the first line of Nalimov's stats is:

wtm: Lost in 0: 12437439

while for me is only:

wtm: Lost in 0: 7844956

The difference in this case is due to the presence of many illegal double checks.
So my stats are not very interesting for an exact comparison as the differences are generally not negligible.

If you are interested also in some 3-3 and 4-2 TBs stats obtained by my program I can send you them in the format of Nalimov's stats but with different names to avoid any confusion. The problem for me is to remember which stats have been already sent to you so in general I prefer to send again all the stats I have in that moment.

Ciao
Guido