Originally posted by cotoi
Is it so easy to find the perfect moves in a difficult ending without using tablebases? GMs make a lot of mistakes in such endings, it seems that amateurs are better at playing an ending.

It depends on how you define "easy"(not my word) and "mistakes". I certainly think it is easier when you have a large number of GM and IM examples from which to cull. Each line is an example to follow or prune, and sometimes you really only need to see the critical or defining move or technique to figure it out. Sometimes...

As for GMs and mistakes, I think we can say "GMs make a lot of mistakes" if we define "mistakes" as "not the most accurate way of proceeding". They don't always do it in the least amount of moves, and occasionally they err, but by and large I think they play endings as we would expect GMs to- very well.

Originally posted by toeternitoe
Missed part of your post.

A 1400-1500 equipped with an engine pitted vs Carlsen.What would his input be other than executing the moves on the board?zero,nul,nothing,nada,zip.
Better to just let the engine run by itself,the human might blunder by moving a piece to the wrong square.

toet.

The point is that programs these days are good enough to beat Carlsen with his inflated rating or any other unassisted GM, the 1500 player would be a kind of "delivery man" or "postman" from the board to the program.

That, unfortunately is how some people play online, they are weak players who just type the moves into the program. There is also the "man and beast" combination featuring "advanced chess" or "centaur chess" which we have talked about.

Whichever way you look at it, can you be really sure that humans can see the endgame as if they had tablebases? That brings us to the discussion of the game at hand.