Author: Ratko V Tomic
Date: 15:17:40 11/21/99
Go up one level in this thread
> I think there is a counter to your argument about this. > > I agree that with a sequence > > a x b y c z d > > that if d is a bad move that only manifests itself over the horizon, then > it is dangerous to have backed that best line and score back to the root. > > If d loses then there is a danger that a loses. > > The shorter the line the greater the danger and the longer the line the > less the danger, since side 'a' can deviate from b,c,d,e,f,g,h in the > move sequence. > > Or, that the deeper the line, the less chance there is that the position at > d is so important, thus causing a to lose. That's all true, but that is the case excluded in my post as well as the "short termination" nodes, i.e. the nodes within the D+1 horizon (but not within the D horizon) which have a decisive advantage (checkmeate or a sufficient material gain), thus they're either known exactly (a checkmate), there is no error interval to overlap, or are far apart enough from the nearest competitors in the up-propagation that their error intervals still do not overlap. Hence the propagation mechanism I described for the nodes differing only slightly (in their sums of the small positional gains, which is what decides the move in the absence of "short termination" nodes) clearly doesn't apply in such cases where the error interval overlap is absent. So I don't think we disagree on this at all. (In the longer message I just posted in this thread, I gave it another shot at clearing up various misunderstandings.)
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.