Moby
Dick
Yeah definitely. As you say there's some degree of subjectivity that comes with these models so a bit more can be adjusted.That's what I'm thinking - yes.
It isn't ONE universal model anyway - there are multiple models, using slightly different metrics.
To me, at least, it would make some sense to adjust for - precisely - "clear and obvious ref errors". So - if you will - I would prefer a model that did adjust for that to one that did not.
Again - because "clear and obvious ref errors" is a genuine thing, i.e. a potentially significant (enough) factor over time.