Great write up man. It sure is hard to make necessarily repeatable tasks in MMOs feel engaging while not feeling like a hassle.
Allow me, if you will, to spin off your idea just a bit. I mean this as a compliment :)
When I think of your idea from a realistic PoV (as some others have done), it seems to me that it doesn't quite track, though there's so much to love about the concept! (Also, I worry that the mini-game involved necessarily leaves my character open to attacks since I'm engaged in an immersive task in order to spot my target).
Here's what I'm thinking:
Just like you point out, a Herbalist should start out with approximately zero knowledge about "plant X" and through experience and in-game knowledge acquisition should improve.
What if we allowed herbalists to self guide their knowledge gain at the point of learning (when gathering the plant)?
Imagine Helga the Herbalist sets out to gather enough Moonleaf to make 10 potions. This will require approximately 100 Moonleaf plants.
What we would like (from a meta perspective) is for Helga's job to get easier as time goes on. "Easier" in this context (IMO) should mean faster to spot, more yield per harvest, better outcome of research. So, plants 0-10 should be really tough to find whereas 90-100 should be much easier IF she stays in relatively the same biome.
So Helga sets out with only a very broad idea of what Moonleaf looks like because she read somewhere that it's generally a silverish plant found at the base of certain trees (or whatever). Helga starts scouting trees, looking for silverish type plants.
Finally, she spots one (using basic visual search). She approaches and harvests the plant. Once that's done she has her plant (though the yield will likely be low - maybe only 3/XX leaves). Now, if she wants to "research" that plant in order to be able to identify it better, she can choose to "sacrifice" it for research.
When researching, a new UI opens with a ?screenshot / 3d model) of the plant - though jittered / blurred / torn / modified
The idea behind altering the plant is to simulate the idea that A) our perceptual system is noisy - meaning we don't have Perfect perception, B) that expertise improves the level of detail one can perceive.
Consider identical twins.. A random person seeing them for the first time (or even the 10th time) will struggle to tell them apart, whereas their Mom does it effortlessly. This is because she is literally seeing differences that the 'novice' is "blind" to (not really blind, but rather doesn't have the sensitivity to - yet).
Then, once that new UI is up, Helga gets to literally click on some feature(s) (say the leaf pattern, or a flower pedal, etc.). Perhaps the number of features she gets to select is dependent upon her Herbalist skill.
Selecting the feature means she's spent time scrutinizing some particular aspect of the plant such that it becomes more familiar to her. Now, next time she looks for that plant, she has a better chance of spotting it as, over time and experience, the plant will "pop" more (perhaps eventually leading to some highlight). Also, that feature becomes less 'distorted' such that when she inspects that plant in the future, she gets a higher fidelity "view" of the selected feature (eventually leading to a true representation of the in-game model).
This seems to track real world performance. Try going out with a professional bird watcher some time... they're like "Robin." "Flat tailed So and So", "Green crested Dingler". while you're like "wait... there are birds here??".
Now here's where some RNG / Skill is involved... it's quite possible that there are many plants that share "feature Y" that Helga selected. Especially in the vicinity of Moonleaf (because spiny leaves are an adaptive phenotype given the types of herbivores in the area). So selecting that feature wouldn't quite help you in that environment until you scrutinize more features (etc.).
The feature list you slowly accumulate could then be scribed into a book (or not). That said, even if I got ahold of your epic herb encyclopedia, it would only help me to a degree... (try reading a book on how to juggle... you will fail the first time even though you can recite the technique). Similarly, knowing that you're looking for X, and actually spotting X are very different. Just ask any radiologist screening X-Rays for tumors.
Anyways, that's my idea based off of yours :)
What'yall think?