I think Stephen’s assessment is sensible - IDs can be modified or obs downgraded as necessary. As time goes on, the data can improve.
If there is a data quality issue, then maybe it should be possible to flag a taxon as one that has overall quality issues and is in need of some expert love. (See below)
I’m not a big fan of restrictions, particularly when you consider how often iNat is used as a tool in classrooms - I think it’s better to change user behaviour with nudges. How about, instead of blocking the agree button, having it so that for the first n agreements a user is required to add a comment about that agreement? Context is a good thing to have generally.
As above, if there are particular taxa that are problematic, and flagged as such, then user behaviour could be prompted in a similar way. (‘Why do you think this is X?’) (PS, see below about guides)
I hear what you’re saying, but I’m really bad for this because I tend to start uploading late at night, and then get prodded that it’s (kiwi) bedtime so end up sending off a few high level IDs in with the intention of going back to a guide the following day to look further.
Finally -
(In NZ at least) There are some really good guides on iNat that could help a new user (or any user) but I think they’re effectively hidden in plain sight. Guides can be far more useful diagnostically than the raw Wikipedia article reproduced on the species page (Compare https://inaturalist.nz/guide_taxa/277342 with https://inaturalist.nz/taxa/57516-Bombus-terrestris, thanks to @tony_wills) but guides don’t seem to appear contextually around the site (unless I’ve missed something). Maybe they should appear where relevant on a species page.
Summary:
Nudge to change behaviour, don’t restrict. Somehow link guides in so users can learn more.