Why do Computers show +0.1

[Comment Deleted]

You could extend that argument all the way to 0.0 or M[x]. The reason engines don't do this is because they can't analyse deep enough in a reasonable timescale. A heuristic is used to evaluate the position at the end of each search path. If Checkmate or a forced draw for the player with the weakest evaluation, isn't found then the number shown is the best from the heuristic evaluations found in the search. +0.1, for example indicates that the outcome is uncertain with best play but White probably has a slight advantage. If you follow the best engine moves to the end then the match probably results in a draw but as the position advances the analysis changes and the outcome might result in a win for either side.

Everything between -0.7 and +0.7 is a draw.

Even in drawn positions there's a stronger side and a weaker side.

Depends on whether the eval gets closer to zero or larger as the game goes on

It just depends on the position.

@9

"Or would the Eval bar swing back and forth a lot"
++ Theoretically there are only 3 possible game evaluations: draw, won, lost.
Say 0.0, +1000, -1000.
All engine evaluations are only approximations, that get more accurate at higher depth.
+0.1 basically says most probably a draw, but a very slight winning probability.
Per Kaufmann: between -0.7 and + 0.7 draw, > 0.7 white wins, < -0.7 black wins
Something like +0.5 can become +0.7 and ultimately +1000 at higher and higher depth,
but can also become +0.3 and ultimately 0.0 at higher depth.

Kauffman worked on Rybka, what, over 10 years ago? It's misleading to quote him.

@14

"Or at higher depth would it find out its just a draw?"
++ 'If you let the computer run long enough it is a draw' - Carlsen

That quote may be old but he still works on engine evaluation algorithms and tuning with Dragon.

@17
More recent quote:
'Statistics of the previous two superfinals show that a Leela book exit of +0.30 or lower is an almost 100% certain draw.' - GM Sadler on TCEC imposed openings

@19
The Top Chess Engine Championship has 50 imposed openings to prevent all draws.

I always just interpreted +0.1 as a valuation of white having first move/tempo

@21
A tempo in the opening is worth 0.33 pawn. Thus oscillation between +0.33 and 0.0 is normal.

Basically yes.

However, today's engines are products of 60+ years of hard work by thousands of chess experts and computer experts. So those estimates are not as rough as you might think.

I don't know how you imagine this, but you are wrong. Positions with 7 (and less) pieces are solved now, for all possible continuations. It is impossible to get out of the tablebases by playing "inaccurate", how would that even be possible?

The other thing is that the engine and the tablebases are two different things. If the engine has access to the tablebases, it is perfect in evaluating those positions, but most engines don't have that.