@25
"When a player is ahead by 3 queens, a move may be called "dubious" even though the player is completely winning and it doesn't matter anymore."
++ Absolute loss of material does not matter. It is often good to give back some material.
What matters are moves that transgress the drawing margin of -0.7 to + 0.7.
When ahead 3 queens the smartest way is often to sacrifice 2 queens to eliminate counterchances and avoid stalemate.
"I think a better method would be something like the classical 5 evals"
++ There are only 3 evaluations: draw (roughly -0.7 to +0.7), won (> 0.7), lost (< -0.7).
"bad moves are only those that change the eval from one to the other (since you can only change an eval down and never up). ++ Yes, that is spot on. So there are only errors (?) that change an evaluation down 1 stage and blunders (??) that change the evaluation down 2 stages.
When you are already down by three queens, losing a bishop may not substantially worsen your chances but it is still confusing to describe such moves as "Excellent" or "Good" when you are losing a piece.
I suggest that English is nuanced enough to replace "Excellent" with something like "Suboptimal" and "Good" with good old "dubious" or "?!."
But what about the other edge case? When a player is ahead by 3 queens, a move may be called "dubious" even though the player is completely winning and it doesn't matter anymore.
I think a better method would be something like the classical 5 evals, and bad moves are only those that change the eval from one to the other (since you can only change an eval down and never up).
But this is not as marketable.
The average ratings are ~800, so most of the customers are beginners. Being accurate / informative will be confusing and off putting. Plus, telling a near-beginner they played with 80% "accuracy" and a bucket load of good and excellent moves is good for business.