I don't know what the average, standard distribution, distribution function, etc... of a bunch of humans scoring targets would be. Running an experiment like what you proposed would probably be a good way to establish that. But you proposing that an etarget that will randomly place 1 out of 20 shots more than .5"* away and between 6 & 7 shots out of 20 more than .25" from their actual location will give the same results every time is just not believable.
Human error and machine error look different, of course. Misplacing a shot that hits a giant stack of pasters is different than just randomly misplacing shots all over the place. But human mistakes are auditable, an e-target is FINAL, even when it massively F-s up. I have seen, and made, my fair share of errors scoring targets, but I've never misplaced a shot by a half inch. I have, and have seen done, forgot to move the score marker or put it on the wrong score, but that is an error easily corrected after the fact, unlike ANY ERROR an e-target produces.
*IIRC the current tolerance offered by e-target manufacturers is an SD of .25"