Ehm... that is not controversial at all... (Although it is obviously wrong to state that TBs do not improve the endgame.)mcostalba wrote:This is perhaps even more controversial than the assumption above, but IMO TB are useful to make the engine stronger in the middle-game, not in the endgame. It's in the middle-game that you lack the depth to workout the endgame at the end of the PV, and it's in the middlegame that TB can more often overrule the search (because at middlegame the search of the endgames sub-tree is very shallow and mainly rely on the evaluation).
Have you really never read anything about this topic?
Is it still so difficult to visualise that probing one time in an interior node and cutting the subtree is more efficient than probing at least once in a leaf below that interior node? (And it is also more accurate.) Probing closer to the leaves certainly does not result in probing less. It is pretty amazing what you are writing here...Assuming the above it is possible that probing at the end of the search is more efficient because you call probe less times (for instance all null search, razoring, etc don't end up in a TB probe because engine cut-offs earlier)
Does anyone else here have a problem in understanding this? Maybe a drawing is needed?
(What may be more efficient is using a higher probe depth for 6-piece positions than for 5-piece positions, knowing that the 5-piece positions are more likely to be cached in RAM. As it happens, this idea was already implemented in SF. But probing the same table in the leaves when it could have been done just once in a parent node is just silly.)
You are probably thinking that something like alpha-beta is the result of lots of random hacking and testing instead of a flash of mathematical insight.So complex that only testing legacy vs natural TB in real games will give us the answer.