My take on this -- based on what I read on Language Log, mostly -- is that there are two classes of grammatical “errors” (though in some cases the distinction may be not completely clear-cut): 1) actual processing slip-ups due to brain farts or similar resulting in phrasings educated native speakers wouldn't normally use in the relevant registers, even when they're using System 1 (fast) thinking alone, and 2) failure to conform to artificial (or artificially-kept-alive) rules introduced mainly for signalling purposes, and which even native speakers don't normally follow unless they're using System 2 (slow) thinking to do so. If you see a pattern in the kind of errors people make (e.g. so many people using “less” instead of “fewer”), then you're likely dealing with an error of the second kind (unless there's some other explanation for the pattern, e.g. it involves identically- or similarly-sounding words). (In the British National Corpus, “fewer” followed by a plural noun occurs 625 times, whereas and “less” followed by a plural noun occurs 162 times. In actual processing errors, the ratio between the frequency of the correct form and of the wrong form would be much larger than that -- for example, “these” followed by a plural noun occurs 29603 times, whereas “this” followed by a plural noun occurs 259 times.)

I can't speak for others, but “actual” errors usually jump out to me even if I'm not using System 2 thinking, whereas violations of artificial rules don't unless I'm looking for them.