why's that, is there some guarantee they fail to live up to? Is there something probabilistic to them or something?
AFCIs are a whole different thing - the plug is trying to predict when arcing is happening somewhere downstream of it and it has very little in the way of 'processing' power to do it, so generally there is a classical filter that is designed to detect the characteristic harmonics of the current waveform that are inherent to an arcing condition. When you experience a sharp upward or downward step in power draw you introduce harmonics, when you have capacitive coupling between SMPS you introduce harmonics, motors often introduce harmonics, etc.
There's like a zillion things that can create AFCI trip conditions and cause spurious trips and frustration for users, and there's very few documented cases of them doing the thing they're supposed to do. I think it's a case of someone being well intentioned but releasing regulatory guidance to use a product that isn't quite there in terms of technology maturity.
Arc detection is also basically somewhat complex RF processing, so there are multiple approaches to it and implementations vary.
Plus (detected) arcs are common with for example brushed motors.
Maybe there's something else the person you replied to would like to highlight, but I wouldn't call them a money grab, just made difficult due to some past choices.
According to the CPSC, who analyzed some 10s of thousands of fires, something like 50% of them were caused by wire arcing that could be prevented with AFCI breakers. This is the exact reason why the NFPA is moving the NEC towards eventually requiring AFCI on every circuit in the house.
Do AFCIs actually do their prescribed function effectively?