question about quiescient search and static eval

[jaespo]

If a search algorithm uses quiescent search, won't all of the leaf nodes have much of the pressure in the position relieved?

(Definition: pressure in this context means threat to capture)

Wouldn't that make the static evaluation function inaccurate, because the amount pressure in the position should be a part of the evaluation?

[xmas79]

The amount of pressure is indeed included in the evaluation. King safety and threatened pieces are two examples of evaluating the pressure.

[hgm]

The point is that if a threat to capture is worth more than actually performing the capture (because it will turn out to be a trade, and leave a position without threats), the side to move will prefer to stand pat (harvesting the threat-to-capture bonus in the current position) rather than trade.

[jaespo]

Ah, I knew I was missing something!

I just took another look at the code in https://chessprogramming.wikispaces.com ... nce+Search and I see now that it calls Evaluate at start of each qsearch node and uses that as a stand-pat cutoff.