Interesting. Am I to infer that if I were to drop a Q whilst playing K+Q+R v K then this would not even be classified as an inaccuracy as I would still be winning easily?
I watched a friend play a game on Lichess recently. It started 1.d4 Nf6 2. c4 g6?! (this being tagged as an inaccuracy). A Benoni position arose, W lost a piece to a pin down the e-file, swapped off into a losing ending and played on until mate. Both scored relatively lowly in terms of ACP loss despite my friend playing a “good” move that dropped the assessment from +61 to +24
Thus there would appear to be difficulties in using inaccuracy/mistake/blunder/ACP loss stats, particularly from one game, but over a 7 round tournament, I would have thought some correlation might become apparent.
I think one issue is this information is so new to us we are still unsure how to interpret it.
For example, I’m not sure what Mr Thomson is saying (with regard to engine matching). Also, we have probably not given much prior thought to how OTB cheating manifests itself on the board (as opposed to at the board or away from the board) so we are a little in the dark as to how to distinguish online between the rapid improver, the occasional cheater and the inveterate cheater.
Certainly good players make lots of errors (one reason they are not 2800). Are there “tells” exhibited by cheaters? For example, one might expect them to be suspiciously accurate in complex tactical positions but play odd moves in endings? Time taken might be a clue. What happens when two players using assistance meet?