My point was getting a more accurate result without the fluctuations of early evaluation of shallow depth.
Obviously the evaluation with finite depth is just an approximation, but depth of 1-5 is very inaccurate as opposed to higher. What I'm suggesting is the bar not updating until a certain depth is reached.
So let's same on move 12. the evaluation is +5. On move 13. it's +5,2.
But going from move 12. to 13. in self analysis the bar will jump from +5 to something like +8 and then settle at +5,2 and it would jump like that on every move.
My suggestion is to freeze the bar until a certain depth is reached, so that going from move to move wouldn't result in the bar going all over the place.
That way when it moves (after waiting a bit) it will more accurately display if there is a change as opposed to it jumping all over the place it'd be harder to see if it's better without memorizing the numbers.
When analyzing a game without the premium analysis, there is only the computer evaluation bar that shows which side is better. It does however tend to jump around a lot. For example going to the next move, it may jump up to +7.9, then adjust to +1.4 as it increases the depth. This makes it rather difficult to see at a glance which move made your position better or worse, as the position might be 1.4 for 10 moves, but as it keeps jumping each time the move is changed, it's hard to tell.
Maybe there could be a sort of delay on the bar, that would keep the bar from the previous move until at least 50% of the depth has been calculated so the jumps between moves would be more reflective of the actual evaluation changes.