I’ve been enjoying the push to get more people to add identifications on the platform and see this as important way to grow, improve, and sustain the community we have. However, I have started to wonder about the possible downsides to adding more identifiers. It definitely has the potential to exacerbate some of the problems we already have with leaving unresponsive users leaving bad misidentifications behind. If too many new users try identifying and then leave, I could imagine a significant increase on the strain in the system. I’m curious to get other’s thoughts on this. In recent years, I’ve focused less on commonly observed and identified species, so don’t know if this problem has generally gotten worse or not.
In particular, I’m curious about the following:
Have you noticed an increase in the number of tags to help overturn bad identifications?
Have you noticed generally more bad identifications than in the past? This is probably quantifiable by seeing if the number of Maverick IDs is on the rise.
Have you noticed an increase in observations stuck at a higher level taxon due to a combination of how few people identify the organism and the increase in bad identifications?
If there are other potential downsides, feel free to discuss, though I’d prefer to steer away from the discussions that have played out many times over on this forum. We all know that it’s hard to get rid of bad identifications without tagging multiple people, it’s just a question of whether this exacerbates the problem or not.
it’s probably hard to do just based on anecdote, and it probably would be difficult for regular individual users to get representative statistics, but it probably would be best to consider increases in those statistics / trends that you’re most interested in relative to increases in observations in general.
You’re probably correct. I do wonder if there might be an increase in users who don’t stay long, though. I don’t know. Just something I’ve been wondering about today.
I’ve been on iNat for about seven and a half years, I do a lot of identifying, and I haven’t noticed any increase in bad IDs or the like. On the contrary, I think the iNat computer vision/AI has improved so dramatically for my region (northeastern North America) that I think the number of bad initial IDs made by observers has gone down substantially.
If someone is identifying in good faith and shows any sort of aptitude it is only for us to perservere with their mistakes, and supportively help them to recognise the species they are struggling with. Then they will become the top identifiers of the future. Some will fall by the wayside quickly, but worth the investment overall.
the rate of bad IDs I see is low enough that I think at the present moment, it’s 100% worth it to get more identifiers. Unless the community ID is stuck at life or it’s a rarity there’s only a few observations of on the website, which is, well, rare, I think having more people going through insects, plants etc would be very worth it. Maybe if they had an error rate of like 10% I’ll change my tune but I’d expect more like 1%.
Either way, more identifiers in general also means more identifiers that stick around and are or get good at ID’ing, and then help overturn any bad IDs that unresponsive users might have left.
I see wrong / abandoned IDs because my filters are set to look for them. Wouldn’t say I have noticed an increase. More that I appreciate when I see an active (and competent) identifier to add to my @mention list.
I would like the CID algorithm to be less vehement. 2 wrong IDs should be convinced by 2 ( = RG) plus 2 more to counter the wrong ones. Let us off hunting down a FIFTH identifier.
If you set iNat’s stats to 3 months, that includes the CNC spike, the IDs level off a bit higher after CNC (more obs = more IDs) https://www.inaturalist.org/stats
I noticed a young identifier recently take over where I had left of with a specific under-identified genus a few years back. They initially tagged me with questions on a few observations, where I gently pushed back on overconfident IDs and suggested some resources and improvements. I haven’t been tagged as much recently, not sure if it’s because they’ve been more confident or less active or if I was less responsive than they hoped. I haven’t looked through all their identifications for quality control; I’m sure I would find some mistakes as with anyone learning a new taxon, but in general I don’t have time and think it’s better to have someone identifying that taxon rather than no one.
It’s kind of nice where you’re the only identifier for a taxon because you have full control over where every observation is going and they’ll all be identified to your confidence level. But that’s not sustainable with the growth of iNat so as long as people are moderately reasonable and respond at least somewhat fairly when mistakes are made, it’s great to have help.
I mostly stick to identifying just a few taxa, mostly at the species level. I haven’t noticed more bad IDs, but then I haven’t noticed new identifiers for “my” taxa either, though I would welcome them! Thanks for thinking of things we can quantify and watch out for, @nathantaylor.
I’m not sure whether I’ve seen an overall change, but I did notice an increase in wrong IDs of ‘edge case’(?) taxa around the time of the monocots/dicots ident-a-thon. From what I saw of the materials, it oversimplified the differences and ignored the ‘but then there are…’ examples, leading to quite a number of ‘by the book’ IDs that happened to be completely wrong. I haven’t followed any of them up to see whether (a) people withdrew their wrong IDs, and/or (b) the confusion discouraged potential new identifiers, though.
I’m glad to know no one has really noticed a general increase in identifier inactivity at present. That gives me some optimism that it won’t increase much over time.
@grampianshiker That’s an interesting note about edge cases. That’s something I’ve generally noticed an increase of as well. I’ve also been a little dismayed by how few have withdrawn their IDs, but some of these are power identifiers.
Perhaps future training endeavors should focus on how to handle disagreement and what to do when another user disagrees with your ID. This is probably more important than learning how to identify in the first place. This is especially true if you’re on a bandwagon that’s holding up a better identification.
Just to follow-up, there are many specific issues with starting to identify on iNaturalist that has become second nature to power users (so much so that I don’t think about them much). These include things like:
Techniques for cutting down the number of notifications you recieve.
How to manage being overwhelmed by increased notifications.
How to deal with uncertainty when your ID is questioned.
How to determine whether the other identifier knows enough to make the ID they did.
How to assess one’s own level of expertise.
How to determine the observation in question is beyond your current level of understanding.
I don’t know how much training on these items went into the ident-a-thons, but these could be important items to focus on moving forward. At the very least, perhaps a little information on the importance of follow-up combined with offering a more advanced course on these topics.
How do you manage your notifications/ID’d observations? I try to go through every notification to check if someone is disagreeing with my ID but I am wondering what the workflow of other identifiers with more experience looks like.
Also, I noticed that as soon as I open a notification, all other notifications are marked as ‘read’. Does the same happen to you?
That’s a good question! But a little beyond the scope of the topic here. Would you be interested in creating a new post for that one? You’re welcome to link back to the comment when you do. My methodology for this is rather involved.
Identifiers who do not get guidance will make a million misids. I can’t help because I cannot work on infinite guides at once. A few conveniences:
Make and link that guide taxon on even 50 observations. No one will ask you for explanations, and you would not be copy pasting 5 paragraphs for each id.
New identifiers can refer. Tags can help them sort the species out. No frustration of correcting a new identifier a thousand times. Fixing wrong ids from new identifiers takes quite some of my time, and may take yours too (unless you are a generalist
Correcting misinformation on the web. See comments here and its replies, especially teh one on ID sections.
I’ve not specifically noticed an increase in mis-ids. But in somewhere like Turkey the main users are European/Russian and they’ll commonly ID with whatever the European equivalent is (aided and abetted by the AI), and that will often be seconded, whereas of course in Turkey they are quite regularly different species not found in Europe or niche species that are not the common Turkish one. Given the small finite number of those with skills in Turkish plants, more identifiers, if not chosen for their experience of Turkish plants, could easily make things harder to get a plant to reach its correct identity.
If you can get your Endemic Turkish plants across the 60 obs threshold, then the CV will become a help not a hindrance. Cape Peninsula attracts many tourists, who ID to a name they recognise from home – but we have botanists and taxon specialists. We get there.
With the new ID tips, you can nominate and upvote your comment – Turkish Endemic because …
Many endemics would be very hard to get to 60 observations, there are some that might get into double digits but others just 2 observations would be an achievement.
I’ve been following the ID tips discourse but, although it’s better than nothing, I think it’s been designed entirely the wrong way so I’m not currently planning to use it. I might relent; but I think it’s a huge missed opportunity :)
