Talk:Endgame tablebase

Questionable Attribution in Background
In context, the following paragraph seems to have an error:

"In 1970, Thomas Ströhlein published a doctoral thesis[11][12] with analysis of the following classes of endgame: KQK, KRK, KPK, KQKR, KRKB, and KRKN.[13] In 1977 Thompson's KQKR database was used in a match versus Grandmaster Walter Browne."

Contradiction: "generated by retrograde analysis" vs "Step 1: Generating all possible positions"
There is a contradiction between that statement from the introduction,
 *  "Tablebases are generated by retrograde analysis, working backward from a checkmated position." 

and "Step 1: Generating all possible positions" which considers positions that are obviously not checkmated positions (which require the enemy K to be on the border). So I wonder which one is right. I do think that retrograde analysis should be the way to go. Then, for many standard "mates" [K+Q, K+R, ...], one can restrict the search to the case where the opponent king is on the border -- i.e., only 4, not 10 squares, and clearly much less positions of the mating material: one of the pieces must threaten the king; and if it is next to it, it must be protected by the own king or another piece. &mdash; MFH:Talk 21:56, 3 February 2022 (UTC)


 * While I agree that the searching principle(s) may be somewhat unclear in the article, I do not immediately agree that the ending positions can really be restricted to checkmated positions. A complete retrograde search would need to include stalemate positions, but also... any position that can be a repetition of an earlier position.Elias (talk) 14:43, 15 February 2023 (UTC)

ÑÑ The most basic utility of endgame tablebases doesn't seem to be treated anywhere. If some piece is a statistical traitor, wouldn't that be something that human players should try to figure out? - Joshua Clement Broyles ÑÑ — Preceding unsigned comment added by 186.28.50.13 (talk) 17:24, 25 May 2022 (UTC)

Metrics are outdated
DTZ metric is more popular than DTC and DTM and should be mentioned in bold. Can anyone think of a simple position in which the difference between all 3 metrics is highlighted? Jack234567 (talk) 07:05, 31 October 2023 (UTC)


 * I've added some information about DTZ but we still need a better diagram. Jack234567 (talk) 04:08, 1 November 2023 (UTC)