I pick analysis because I think that is more beneficial to (most) users and I think a natural by product of improved analysis would be gains in ELO but I don't know that it would work the other way around.

As an engine user I favor quality of analysis over anything else. But I don't know of any better objective measure of it than Elo rating at intermediate time controls (like 40/20' or 15' + 15" or such). Of course I have subjective opinions on the issue. So in Komodo we aim for quality of analysis, but we can only measure it by intermediate level elo.

Consider this Analysis over enlargement! More specifically! Will this boost in enlargement affect her game play in analysis with on court concentration. Or, give her an added boost of self image confidence apparently lacking before the enhancement. This could just leave her on her back!

I would say Vas still has a reasonably strong market if he continues to advance a chess engine strong in analysis. At the same time Houdini with its advancement in elo has little to worry for the time being. Actually, I think the field is wide open with Komodo possibly showing some new advancements in analysis along with Critter.  In the long run elo strength may not be the safest niche to occupy.

Judging from previous polls there might be at best 10 more member votes out there to decide this issue before it is closed.



D-cup lovers might consider this wonder product- The   "HOUDINI EXE-O-MATIC ELO INFLATER "

         Houdini exe-o-matic elo inflater

(Ummh?! Role model issues presented here!!!! These two birds don't look Japanese! Oh! Heck! Houdini isn't really-well, a commemorative of Erik Weisz either! eh! )

This is the same argument that was going around pre-fruit and especially pre-rybka 3-4.   Indeed, it was felt that we had hit the wall in engine elo development, and that analysis, and...  Human Bad Playing were the next frontiers of chess analysis.

When there was a substantial gain in ELO, such as a full class difference between the previous front-runners and the new guy, that the change in class represented a closer approximation to the "truth" of the chess position.   In that particular case, the answer was way more important than the analysis of the answer.  Now, modicums of improvement, (though several 10's of points of ELO is NOT insignificant), are bettered couched in analysis to provide better understanding of the process and validity since we can no longer simply depend on it being the only unassailable "truth".

I think the answer is, I want an engine that is surprisingly and class better than the previous guy.  Something that parts the sky and reveals things we just weren't really doing as correctly and efficiently as before.  Something as surprising as Rybka.  I want this more than anything...

In the absence of this, and it actually may be very difficult to do from the engine standpoint, is better analysis.  The problem with better analysis, is that I want it presented so that *I* can understand it.  And it is very difficult to get there.  What I want from a chess engine analysis, is to be like an Andrew Martin chess base video.  That in a given position, I can have the underlying ideas of the position, underlying strategic ideas, underlying threats, and ultimately the tactical truth of the position.   So that eventually, *I* can see it, and ultimately, that I can learn from it and put it into my own games.  I want to become a better chess player.

So first, the truth, second the teacher.