Computer Chess Club Archives


Search

Terms

Messages

Subject: On 'Reported Depths'

Author: GuyHaworth

Date: 16:24:19 08/21/02

Go up one level in this thread



Agreed.  The maxDTM wtm 1-0 KQKRP position has DTM = 104, i.e. it requires 104
winner's moves from the given position.

A search engine probing the btm KQKRP EGT after all the White moves will only
get a maximum figure of 103 back.

It looks like some engines forget to add the 1 move ... or by convention, quote
the depth after the next move - which personally I think is a bit strange.  It
is more an engine/EGT_server issue than a GUI issue.



A caution when using Ken Thompson's DTC(onversion) EGTs on the bell-labs site.
He quotes the depth in winner's moves from the _next_ Black-to-move position.
In an extreme case, this is after 2 captures, the first by Black being forced.
In this case, the count of men on the board and DTC depth of win reported can be
quite surprising.  However, it is usually '1 light' in reporting DTC depth.

The EGTs tend to assume that the winner makes the conversion, and when using the
DTC depth as one too high.  This, and being absolutely precise about 100-ply
depths, cannot be 'sorted' until depths are measured in plies.



I don't know the Chessmaster EGTs, but assume that they are not to the
DTZ(ero_move-count_move) metric, i.e. to pawn-push, conversion or mate.

It would be less than rigorous to infer that White cannot zero or get-zeroe'd
the move-count in 50 moves by looking up a non-DTZ EGT such as a DTC or DTM EGT.

g




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.