I’ve worried about this very thing: the development of local ID cultures.
I was the first big user in my region. And despite being totally amateur, i noticed that once you get to a couple thousand observations or so you automatically get the benefit of the doubt. It’s a real hazard, my ignorance getting propagated.
I see EXACTLY the same thing happening with the Robotic Misidentification Engine.
5 Likes
Incidentally, this was the case with one of the prior paper wasp identifiers back in the day. The IDs weren’t intentionally malicious or anything, but the accuracy was maybe 50-60% (there are about 3 major sets of similarly colored species). This user’s IDs had been taken to heart by many other users as if they were expert IDs as well as a major factor in inaccurate Computer Vision suggestions. This is, by the way, a taxonomic group with several thousand observations, so I want to say it took much of a year to correct. The fairly good news is that the Computer Vision suggestions have become a lot better, there are now a few better resources publicly available, and a few other users have been getting a bit more comfortable (and accurate) in separating these similar species. (There have been a few issues with a curator in Mexico doing the same, sometimes even with species that look nothing alike, though at least most of these are getting corrected quicker than in the other situation.)
A side point is that “seen nearby” isn’t always the issue. Xanthocryptus novozealandicus is a New Zealand endemic ichneumonid wasp. Somehow, it keeps being suggested by CV throughout the Americas with no observations here. There have been some similar threads regarding species ranges and CV that may also be of interest
1 Like
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.