Read on Twitter
I'm a big fan of iNaturalist and use it often, but with so many observations being uploaded now, combined with the auto-ID generating bad IDs (for millipedes, at least), it's becoming unwieldy to use.
Lots of millipedes get identified to an incorrect family based on what has already been identified. Generally, US stuff is accurate, but it can't be applied to other areas of the world.
Incorrect IDs accumulate and get lumped in with the correct ones, giving some wacky ranges and causing confusion. And at this point I've done 15K+ identifications of mostly millipedes, but it can be a huge timesuck to fix mistakes.
Having the auto-ID incorporate geography would help a bunch with this. For example, this genus is only in California, but many pale millipedes get auto-IDed incorrectly https://www.inaturalist.org/observations?place_id=any&taxon_id=122293. I keep cleaning it up, just to watch the boulder roll down the hill again.
At this point, I've mostly stopped correcting IDs I come across outside of North America due to the volume (I can't imagine what it's like for, say, butterflies), it's just too much. Even when I spend a half hour or so each evening checking them.
I don't have any ground-breaking revelations here, but I hope the auto-ID can be improved to somehow manage all the new data coming in. Otherwise, it's going to lose a good chunk of its usefulness. The quality needs to match the quantity.
And a reminder that so much of the curation is done by expert (unpaid) volunteers, who are human and can get overwhelmed.
