This feature request is intended to aid the independence of identifications being made by users and improve the accuracy of identifications. The idea is based on discussions in the “Overzealous Identifications” thread and others related to the use of the “Agree button”.
Users would not see names more specific than Genus on other user’s observations until Research Grade is achieved.
The Agree button will add an identification for the the displayed name, genus or higher taxon.
Before Research Grade, all species level or below IDs require the user to type the name. (autofill options can remain as coded now)
You can always see your own ID at any taxonomic level.
Search results would still have to use initial specific IDs to satisfy certain workflows. Can’t find out-of-range species IDs or uncommon look-alikes if you can’t search for a species. (added after Comment #15)
This may slow down IDs slightly. It may also mess with the ID ratings since the order of IDs matters and also there are flags for what kind of higher level taxa disagreement. Mostly I think a blind system (a little blind anyway) may improve a few issues that iNat is having. Seems like a discussion worth having.
This might be a bit of an understatement, given the top 100 observed species, like mallard, honey bee, human, dog, yarrow, common milkweed, etc. comprise nearly 10% of all observations, and it’s far quicker to hit “a” to agree than manually type in/select/enter the name.
Interesting idea though! Curious to see what people think.
This strikes me as quite a drastic ‘overkill’ solution to the problem of the Agree button. If it was implemented it might lead me to completely stop using iNat.
Overzealous agreement is definitely a problem, but this suggestion would interfere with far more than just that. The first issue is simply that it puts way too much emphasis on the importance of research grade. That’s not what iNat is for! It’s to get people to engage with nature. If a person makes their first observation hoping to find out what species they saw, and the site deliberately hides that information from them, it’s the opposite of encouragement.
There are also practical reasons why it’s really useful to be able to see others’ species-level IDs. The most obvious is that in some taxa and places, there are so few identifiers that a single ID might be the best that an observation ever gets. Even good and theoretically identifiable observations just don’t always get to research grade. Such an observation would be ‘trapped’ at genus level forever, even after a dedicated identifier might have put the work in to be specific. Or perhaps more likely (and more unfortunate), identifiers in obscure taxa might lose interest or give up, if their efforts are likely to be kept invisible - and similarly, observers might stop bothering if they know there aren’t enough identifiers to be sure of hitting research grade.
Even in the context of observations which will eventually reach research grade, there are serious downsides to this suggestion. Seeing what species someone else has suggested can be really useful for making an identification of your own! A genus might have hundreds of species in it, and for a non-expert that may be discouraging or impossible to narrow down further. But with a suggested species to start from, it’s possible to research identifying features, native range etc, and come to a conclusion about whether it is (or isn’t) that species.
I’ve described all these downsides, but I also just realised that there’s a very obvious workaround. Identifiers would simply be able to write their species ID into the comment box. This would be an annoyance and an extra effort every time, but would also invalidate this feature - everybody loses! The end result would effectively just be that there is no Agree button on species-level IDs, which is a much less drastic suggestion that’s already being discussed elsewhere.
Interesting idea! I don’t think it’s quite there yet, but I like the concept of encouraging more independence of identifications. That would solve a good number of ID problems.
Perhaps some of this could be better solved by tweaking the algorithm behind the ID scores for users. Agreeing with someone else’s ID should be worth a lot less than being the first to suggest the correct ID, that being the ID that subsequent people agree with. At least that might solve the problem of identifiers making lots of agreements to drive their names up the ID leader board. I suppose there could be a version of the Identify page where identifiers can’t see others’ identifications, and they get more points for entering their independent IDs there than they get for agreeing with a species name on a regular observation page.
Independent IDs are much more valuable than agreements so maybe iNat should be scoring them that way. How best to do that is certainly worthy of more discussion.
Note that there are many valuable observations that are not and never will be “research grade” (eg captive/cultivated observations, observations without photos/sounds). I’m completely against hiding the species names for all of those. That would be terrible. Similarly, there will be awful photos of things that the observer got a good look at and made the correct ID for. We’d want to see those species IDs also. Let’s definitely not hide the species IDs on regular observation pages.
A variation that might be better would be to only show down to genus level on the identification page, making it so you do have to click on the observation to put an ID on it, whether that is agreeing with the previous person or putting something new in. This would make it so you can’t just agree with everything but would also slow everything down.
If the only ID counting for the top identifier rankings was the 1st correct one for each observation, the agree button would not be a problem. Or… when an observation reaches RG why not to “close” (just the agree button) that observation. If someone disagrees with the ID and it goes back to “needs id”, it would open again until once more the RG was reached. It would stop people from agree with easy IDs just to climb up at the rankings.
The problem I see here is not with the “agree” button per se, the problem is to “identify” something that is already well established as RG. There are multiple users doing that just to go up at the rankings.
This a point I had not fully considered. How to use Obs with only 1 ID? As a user IDing things, I still think it’s best to hide species level IDs until consensus is reached. But as a user that wants to put the data to work… hmm, exports (which is how I use the iNat data) could still provided all the detailed IDs. Heck, I wish the export could do that now instead of just the leading ID. Or separate interfaces for IDs. Don’t make ID a one button process when users are just wanting to browse their data and check for a name.
This is my goal too. Engaging means learning. IDs can be hard. There aren’t enough people that can ID certain things. Having a system that makes it look otherwise is dishonest. I think budding naturalists should know these things. Maybe it will encourage folk if they know that they could become one of the world’s experts of a group. They can. We have plenty of need. Doesn’t even have to be that obscure. How many people know the grasses in their region? Or ants? Or lichens? Almost anything outside of the charismatic megafauna.
I am fully supportive of aids to IDing. Lists of species nearby, look-alikes, and such can narrow down those long lists of possibility. I knew I was understating the “slow down ID” part. I guess my philosophy is that IDs should be deliberate and thoughtful. Not accidental via a thank you or following along.
No, not a complete plan. Don’t let me break the system. But maybe the smart people that got us this far can pull a few ideas out to improve iNat a little.
I do IDs for several taxa that have some very common species and some rare ones. When one of the unusual species is reported, I look at it right away. If all I could see was the genus name, I’d probably miss most of those, as I don’t try to look at every ID to genus.
I often correct IDs that are one the “wrong continent”. I regularly search for a handful of species where Computer Vision and Identifiers make mistakes. Not seeing initial IDs would make this impossible.
Yes, this is a problem. Search filters would still have to use specific initial IDs for many people’s work flows. Once you have your collection of Obs selected, you may be able to ID the oddballs you’re seeking without the names displayed since you know what you searched for. I’ll have to try a few experimental searches. (modifying original proposal for these concerns)
This is good. If nothing else, I’m learning about work flows different from my own. I also know why User Interface design can be so tricky.
I think to some extent we already have this, non-RG observations have a different map symbol and can be filtered out. In terms of overzealous agreement, i think the best answer is to use creative means to encourage people to look at observations beyond their immediate ‘community’… as iNat gets bigger silos are kind of forming, and trying to get a third ID on a section of observations, ideally from people who are familiar with the species but not part of our immediate community, could be really valuable.
My favorite idea is to have an ID mode where you are usually shown needs ID observations but maybe 1 out of 10 that is shown to you is research grade already. You could hide the fact that it has research grade even, though that has the risk of hiding comments as well.
I participated in the blind ID study a while back and my personal observation was that my ID accuracy was lower when I did not have all the info, including what the original user thought they saw, which is useful even if wrong.
A full on mandatory mode where species is obscured doesn’t seem a good solution imho, especiually considering that in some taxa, genus or even family may be hard to ID too.
I really agree with the “useful even if wrong” part: seeing how others misidentify something sometimes helps nearly as much as a correct suggestion- because it causes you to focus on the observation in a different or keener way than you might otherwise.
Thanks for the review everyone. I think the only real useful nugget was stated best by jon. My proposal does not feel like the path to a better system though. I consider this thread complete.
Ok, I’ll close this. If folks want to continue related conversations, check out some of the previous discussions on the agree button / identification / data quality, and if there aren’t any where your comment would be a good fit, there’s always the “reply as a linked topic” option to open a new one:
then - don’t count Agree for the leadership board.
Only count the actual identification.
Which would make the leadership board useful to indicate who to ask for help with an ID.
