It’s always worth reiterating: iNat doesn’t exist for data purity, but to help people connect with nature. Bad data is not a technological problem- it’s an anthropological problem. Each time you kindly correct an identification, you are helping the community grow in knowledge, moving the dial on the nature-human connection.
To that end, it is helpful to offer diagnostic explanations for your corrections. When I go a-campaigning on some genus or species, I’ll take the most common diagnostics and copy them in a document. That way, I just have to copy and paste the relevant explanatory note for the situation, and it takes an extra 10 seconds, that way the user and anyone who views the observation can learn from it. Users in the area will often start reaching out for your help, and suddenly iNat becomes a potent teaching platform.
Don’t be too distressed over error. Enjoy the process. It’s a beautiful thing. Thanks for your thoughtful contributions!
I have decided to go with linking to a journal post I’ve written up. I’m glad I posted. Hopefully I can keep making these for difficult species groups.
I am always amazed at how good the AI computer vision has become! Just in the past couple of years, I noticed it is much better at recognizing images.
For some species, a single photograph probably will never be enough (well, probably someday). For example, how to train it to see a tapered banner petal on a Lupinus bicolor as opposed to the square-ish one on Lupinus nanus. Or maybe it does learn this with enough correct observations with good photos? Perhaps eventually this site will be able to use multiple photos from a single observation entry to process the identification!
This should be only used rarely in my opinion. Because, even if we think there is only something microscopic that separates species, or they can only get identified by DNA, the people who are doing that may discover that that isn’t really the only thing that can separate them and that there is something more obvious. (I’m talking from a Fungi standpoint where you can’t go a day without finding out something new, and I can’t really speak to other organisms.)
Actually, I think, “No, it’s as good as it can be” should be clicked more often! I’ve looked through a lot of older posts, and believe me, there are a lot of 10-pixel birds on wires, anonymous spring leaves in the lily and asparagus families, and blurry grasses that will never be identified. The good in-focus photos that just lack the technical characters we need are a potential problem because, as @fungee says, maybe in the future people will figure out other traits that can be used. However, I think a person who figures this out will (or should, anyway) search on the genus and find these observations.