This is a great question. And while I’m sure it has been discussed before it merits more discussion.
I would wager that there are a significant number of research grade observations similar to what you describe. The issue here is a tradeoff between speed and accuracy. On the side of speed, one could argue that developing a body of RG observations quickly is useful because those RG observations then become an important source of information for other users who can use them to become good identifiers.
On the side of accuracy, I am certain that experts make mistakes. And what good are RG observations if they include errors? If they are honest with themselves, I’m sure experts would welcome having other observers provide careful second opinions about the identity of observations. The entire scientific review process is built on the idea that truths are objective and revealed through careful consideration of the evidence–in fact many papers are now reviewed blind, i.e. with their authors’ names obscured so that the reviewers aren’t unduly influenced by their reputations.
This topic raises some bigger questions about how iNaturalist challenges the historical paradigm of taxonomic expertise. Science is a collective project that requires a community of practitioners who share a set of criteria and who communicate often. Science cannot be done in isolation. However, professional taxonomists have generally specialized on a narrow range of organisms, and shared their work in niche publications that are not always widely read and that are often difficult for others to find–even if you have access to a world class university library. Specialization is obviously a necessary aspect of taxonomy, but the risk is that if everyone has a narrow specialization this reduces the number of other people that can review each others work. And, as I keep repeating myself (sorry!) that’s key to doing science. In fact, when two experts work on the same taxa, they quite often end up getting into big arguments! (Ask anyone who knows taxonomists).
I’m excited about iNaturalist because it opens up this process to a much wider range of people from the beginning. Experts have an obligation within this framework to justify their identifications with objective criteria, just like everyone else. While we can learn a LOT from experts–I don’t want to discount that (I LOVE EXPERTS don’t get me wrong!)–learning from experts is different from simply deferring to their authority without critically reviewing the evidence. Experts can sometimes learn a little from us non-experts too!