Why is the evaluation bar on these live games so different from the eval bar on my phone? The position on live game says “white is demolishing black” and on phone says “white just has a pawn more than black”. Is it just a matter of processing power? Does it really matter if engine is like 3500 vs 3700?

At the limit, there are actually moves engines don’t consider, 25 moves down the line; this is a processing power thing
I also expect multi hundred gaps in Elo to matter even at 3.5

Can lead to funny situations with an engine when you’re playing…..

Be specific.     

leonasskau.co.uk //
Hosting /strategy, /geopolitics d33m:mun

And then, if you’re doing this in analysis, if you go back to the original situation, some engines remember this move, when they didn’t get it the first time round

engine thinks it’s a draw, and you play a crazy move, engine thinks you threw the game, and then opponent plays the engine move, engine sees a few moves down the line, and realises omg legend and swings behind you…

Presumably. It definitely can be the case that even one copy of the same engine can know something that the other version doesn’t, let alone two different engines

Interesting thanks! So basically a big event like this will rent cloud compute to do major processing to get the bar we see on screen. I also assume that for a game like this (between world #1 and #2) you need to look far ahead whereas for my own noob games the phone has a pretty good idea of who’s better 😅