"Universal Grammar" means "universal to all humans". If the grammars of two individuals differ, the difference is in the grammars of the languages, and not in the human faculty of language.
Nobody has direct access to "grammar", we only have access to behavior which is caused by grammar and "other things". People who speak English frequently have different grammars, and we can talk about systematic differences between the English of Chicago vs. the English of Los Angeles, or Hong Kong, or Edinburgh. Some grammatical differences are at the level of the individual (in which case we call them idiolect differences). So differences can be a result of dialect differences, which need not be just geographical in nature.
A variant of idiolect difference is individual difference in "standard of comparison". This brings us into the real of "other stuff", the fact that introspection about acceptability involves variable strategies. In asking speakers about acceptability of sentences, responses may reflect such differences as answering based on the standard "Would I ever say this?" vs. "Would anybody speaking this language ever say this?" or "Would people understand if you said this". Related to this is individual non-grammatical differences in imagination – can you imagine a context where the sentence might be acceptable? A given respondent might be more imaginative in thinking of contexts; or, in rejecting ~ adhering to prescriptive norms.
For over a half a century, linguists have been struggling with the problem of finding out what is "truly in the grammar", filtering out all of the other stuff. There has been a slow uptick in awareness of the nature of those other non-grammatical factors. One thing that may help is to get data from people without a professional ideological stake in the answer – ask non-linguists, or at least non-syntacticians.
It is true that trained syntacticians are quicker, generate judgments that are internally more consistent, and tend to be more analytic (being able to diagnose the source of discomfort and say e.g. "only if you mean that he ate himself"). Chomsky in 1970 (Remarks on nominalization) has a discussion of Raising and clausal nominalization culminating in the interjection (p. 193) of the distinction between acceptability and grammaticalness, a propos the (un)grammaticality of the structure (*)his criticism of the book before he read it, which I and many other linguists find to be perfectly grammatical. He explains that our acceptance of these structures "results from a failure to take note of a certain distinction of grammaticalness". Of course, the alternative is that his grammatical analysis is just wrong, and there is some technical distinction being captured in a dialect difference. A choice simply has to be made, between quicker, carefully-curated but volatile judgments from professionally-trained syntacticians, and patiently-obtained robust judgement obtained by "controlled study" techniques (field work and experimentation of various kinds).
