Watching the number of unknowns rise during the CNC was quite the sight, I kept identifying unknowns throughout the weekend and watched the number keep rising. I decided to record the data and the peak was 200k unknowns. Below is the number of kingdom level unknowns in hours since the start of the CNC. As of the posting of this, there are still 150k unknowns.
When I first looked at La Paz Unknowns from this CNC, there were 40,000. Last night it was down to 29,000. So people are working on them.
Of course there are many just IDed as “plants” and many incorrectly identified ones which don’t show up in that figure, but are really still Unknown.
I’ve been working through them too. I generally flick through and only add an ID if I can do family or better for plants, and phylum/ order for animals that are less common. Depends on my mood though. One thing I found really satisfying was watching the number of pages decrease as I IDed them. They were going down a lot faster than my humble input, so I could imagine a bunch of people all over the world busily working away :)
And the joy of persevering with clearing those CNC23 URLs. Because. They will. Actually go down to zero. No longer IDing as fast as I can, and still ending up with even MORE to do!
In reference to comments about La Paz, it currently ranks 1st for number of observations in the 2023 CNC …so although I haven’t done any calculations, the percentage of unknowns might possibly be in line, statistically, with other cities.
They are less than 7% of all observations, so it is about 3x the overall rate of unknowns. La Paz obviously has an amazing level of buy-in to the challenge which is great, so they already wipe the floor with everyone! If they were to attempt to mirror the IDing operation of say Cape Town then I can hardly imagine what they would achieve… It is early days for them I guess - they came out of nowhere last year - but I hope that they are able to better equip their observers in the coming years to give IDs but not wild or overspecific IDs. They seem to have the network to do that, I just hope they have the intention. As it stands, their success in generating participation is hard for everyone to manage practically.
I believe that La Paz will generate some fabulous naturalists as a result of this participation - but some will perhaps be missed because their observations are so buried…
There is also not a single journal post published in the La Paz project.
Maybe they used another platform for guidance, but in the project description, they mention their success from last year and that the national and internation communitiy will help with identification.
No link to any guidance, no mention of how to provide good photos for ID, no follow-up after the event.
And all that with more than 1200 members in the project.
I see this as a huge missed opportunity by the organizers.
There are no journal posts in the umbrella CNC project, either. In my local CNC, we tried to use journal posts as a way to engage with project members (all 44 of them), but I don’t think that was particularly successful.
That’s a really good point. If the umbrella project were used to provide a bit of guidance it would take the onus off the local groups a bit - hopefully result in a bit less variability.
True, but that assumes that everyone joins the umbrella project and that they read all the journal posts on the umbrella project. At least for my local project, not every observer joined the local project (I’m sure some observers had no idea the CNC was going on) and even the local project members didn’t read our journal posts. I can’t say I read the journal posts for every project I’m in or person I follow.
Maybe (maybe!) it is better fot learners to post an observation as unknown rather than to make a guess with a unlikely identification or to follow the suggestion by the AI as pure gold.
As regards, this morning an account possibly of a class of students posted almost only ornamental plant with mostly wrong IDs (all suggested by the AI). The question is always the same: what do these student learn from this way of using iNat apart being able to create an observation?
Of course the AI will suggest wrong IDs so long as cultivars, hybrids and cultivated forms are not included in the AI training. How will the AI know it does not know these species? And how will users ever learn, if the AI gives wrong IDs for these?
Am having this problem with the Proteaceae. Lots of people post hybrids, but the AI slots them to species.
Dont blame the AI for these wrong IDs! Or for the incorrect learning when the AI mis-IDs these.
