This chessforum is currently dedicated to the development of Hoffman, a endgame tablebase generator.
The two most important predecesors to Hoffman are the Nalimov tablebases and Eiko Bleicher's Freezer. Hoffman combines the capabilities of Nalimov and Freezer.
Nalimov solves the endgame completely, back propagating from one tablebase to another (for example, KQK is needed to evaluate queen captures in KQKQ), but due to the complexity of the analysis it is limited to roughly six pieces in play. Freezer allows pieces (especially pawns) to be restricted to particular squares and is thus able to solve more complex positions with a dozen or more pieces in play, albeit with restrictions (i.e, it can't solve the endgame completely). The biggest problem with Freezer is that it can't back propagate from one tablebase to another the way Nalimov can, thus it's ability to evaluate exchanges and queening combinations is very poor.
Hoffman combines both approaches into one. It is capable of both back propagation and piece restriction. For example, in the rook-and-pawn endgame we were about to enter in The World vs A Nickel, 2006, it could restrict the pawn movements (like Freezer), and thus calculate how White could go about queening a pawn, while still making a meaningful evaluation of rook trades by back propagating (like Nalimov) from a tablebase solution of the king-and-pawn endgame.
Hoffman was originally intended for use in Chessgames Challenge: The World vs A Nickel, 2006, but that game ended with a win for the World Team, as did Chessgames Challenge: Y Shulman vs The World, 2007 and Chessgames Challenge: The World vs G Timmerman, 2007,
without any use of Hoffman in those gamea. I resigned from the World Team after they accepted a draw in
Chessgames Challenge: A Nickel vs The World, 2008
The program can be downloaded from http://www.freesoft.org/software/ho..., part of my website, www.freesoft.org