Some random thoughts, mostly related to the topic of discussion:
Over my career as a biologist, I used or had a hand in creating several large databases focused on biodiversity conservation - databases like iNaturalist, in other words. None of them were perfect. All of them were very useful, if the targeted users actually spent time putting in and assessing the data. As large databases go, iNaturalist is extremely well-designed, very useful, and highly motivating to use, at least for people like me. Does that mean it can’t be improved? Of course not. But tinkering in any significant way on one part of it is likely to mean the performance of some other part begins to be eroded. If a database is 80% “good,” it’s great in every practical sense, and I think iNat is well over that 80% mark.
With the exception of improving people’s first introduction to iNaturalist, that is. Above, @jnstuart said that, ideally, new users should have an experienced iNat user close at hand. I completely agree. Now, obviously, that can’t happen for every single new user worldwide, but there could be more hands-on Introduction to iNaturalist workshops offered, if we experienced observers offered them to our local networks of birding clubs, libraries, adult education, etc. (I will say that, having conducted a few such workshops as part of the recent City Nature Challenge, this introvert had to go lie down in a darkened room for a while afterwards.)
Not every observation will make Research Grade, which is obvious to everyone reading this, I would think. Some observations should start out as Casual and stay that way. In between, there are many observations that may never be satisfactorily categorized as either Research Grade or Casual, even with the attentions of competent and experienced identifiers. In theory, every observation should end up as Research or Casual, after probably spending some time as Needs ID, but can I say that I’ve gone through all of my own observations (some 38,000+) and labeled as Casual the ones that clearly can never be identified even just to genus or sub-family? No, I cannot say that. Perhaps I should clean up my own Needs ID pile (9,000+) before complaining about other people’s piles.
So, the Needs ID pile is very large and growing fast. Accepting that fact is hard for people like me, who like things tidy and all of the to-do items crossed off our lengthy lists (as if that ever happens!!), but I try to see the good side: I’ll never run out of observations to identify. More and more people are learning something about biodiversity, even if that something is just a tiny bit of what they “ought” to know. More and more good, useful data are available to researchers and managers of biodiversity. All good things, in my opinion.