# Community identification model doesn't work **URL:** \ **Category:** General\ **Created:** [September 23, 2025, 7:59pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608 "2025-09-23T19:59:43Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![malcolmstorey](https://avatars.discourse-cdn.com/v4/letter/m/b5e925/32.png) [@malcolmstorey](https://forum.inaturalist.org/u/malcolmstorey)\ **Post date:** [September 23, 2025, 7:59pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/1 "2025-09-23T19:59:43Z")
**Platform** : W11 **Browser** , if a website issue: Chrome ![numpties](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/5/8/5842923763927394c029853f0ca995d39bd0c0a3.jpeg) **Description of problem** : Some species are difficult and not identifiable beyond genus from photos. If the record is RG at species level, my setting it to genus is just taken as supporting the id when I’m trying to say that it’s not supported by the evidence available. If the record is not RG then I can set it to genus level and set “not possible to identify further from current evidence” (I forget the exact wording, but that doesn’t show on the Identify page and dunno what it does anyway.) so there’s nothing to stop a naive beginner matching it to a photo and giving it a name. I’m working on UK Ichneumonidae. Most photos on the web are wrong. There are only a handful of spp in books. Most records in iNat are wrong. The only hope a beginner has is using ObsIdentfy or [https://waarneming.nl/](https://waarneming.nl/) but it’s a lot of work. Ichneumon haemorrhoicus crassigena - dunno where that’s come from. “crassigena” is not mentioned in the checklist or the latest mongraph on the genus. Ichneumon haemorrhoicus is a British species but it’s pretty rare. Be useful to have a button that just says “no it isn’t!” Some photos are not identifiable at all: be useful to have a button to say “you wouldn’t recognise your own mother in that photo” - but perhaps phrase it more politely! Yes, I’m thinking of changing my user name to curmudgeon! ---
**Author:** ![tiwane](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/tiwane/32/54612_2.png) [@tiwane](https://forum.inaturalist.org/u/tiwane)\ **Post date:** [September 23, 2025, 8:13pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/2 "2025-09-23T20:13:00Z")
As far as I can tell, this is not a “bug” [as defined here](https://forum.inaturalist.org/t/about-the-bug-reports-category-please-read-before-posting/60), so I’ve moved it to the #General category. > [@malcolmstorey](#): > > Some species are difficult and not identifiable beyond genus from photos. If the record is RG at species level, my setting it to genus is just taken as supporting the id when I’m trying to say that it’s not supported by the evidence available. You should see a pop-up when you add the ID. Click on the orange button to disagree with the community taxon and that should change the community taxon most of the time. ---
**Author:** ![spiphany](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/spiphany/32/31332_2.png) [@spiphany](https://forum.inaturalist.org/u/spiphany)\ **Post date:** [September 23, 2025, 8:28pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/6 "2025-09-23T20:28:43Z")
In this case the user has opted out of community taxon, meaning that the observation will always be displayed under their ID. Unfortunately the information that a user has opted out is not visible anywhere in the identify module, but if you open the observation you can see it on the right-hand sidebar ---
**Author:** ![matthewvosper](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/matthewvosper/32/12194_2.png) [@matthewvosper](https://forum.inaturalist.org/u/matthewvosper)\ **Post date:** [September 23, 2025, 8:29pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/7 "2025-09-23T20:29:38Z")
Thanks for working on these, Malcolm (I won’t call you ‘curmudgeon’ 😛 ) Actually, in the example you display, the community ID _has_ been changed to genus by your action (the community ID is displayed on the right) ![image](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/d/5/d5688620d7b3124261b640558edde72d7cedbb83.jpeg) The reason the display ID at the top is still the species is because the user has _opted out_ of community ID. In this situation, the observation cannot become research grade unless the user’s ID and the community ID agree. If you set ‘community ID (i.e. genus in this case) cannot be improved’, the observation will become Casual, not Research Grade. ---
**Author:** ![larry216](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/larry216/32/13112_2.png) [@larry216](https://forum.inaturalist.org/u/larry216)\ **Post date:** [September 23, 2025, 8:41pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/8 "2025-09-23T20:41:33Z")
> [@malcolmstorey](#): > > Ichneumon haemorrhoicus crassigena - dunno where that’s come from. “crassigena” is not mentioned in the checklist or the latest mongraph on the genus. Ichneumon haemorrhoicus is a British species but it’s pretty rare. If you think this is an invalid subspecies, you should go to its [taxon page](https://www.inaturalist.org/taxa/451479-Ichneumon-haemorrhoicus-crassigena) and flag it for curation. I don’t have any knowledge about it, myself, but a Google Scholar search does find references to an _Ichneumon haemorrhoicus crassigena_ Kriechbaumer 1887 ---
**Author:** ![DianaStuder](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/dianastuder/32/43358_2.png) [@DianaStuder](https://forum.inaturalist.org/u/DianaStuder)\ **Post date:** [September 23, 2025, 9:02pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/9 "2025-09-23T21:02:37Z")
That is why we would like Opted Out to be more visible. ---
**Author:** ![matthewvosper](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/matthewvosper/32/12194_2.png) [@matthewvosper](https://forum.inaturalist.org/u/matthewvosper)\ **Post date:** [September 23, 2025, 9:15pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/10 "2025-09-23T21:15:35Z")
Yes! To avoid that “why is this observation still in this search when three people have disagreed with the ID? Why has my ID not done anything either? Ah, better check the actual observation page… yup! opted out!” “Now going to make this casual to save others the confusion!” I also think the Community ID needs greater visibility on both the identify and observation pages, for the few occasions when it’s not the same as the ‘display ID’. ---
**Author:** ![schoenitz](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/schoenitz/32/56_2.png) [@schoenitz](https://forum.inaturalist.org/u/schoenitz)\ **Post date:** [September 23, 2025, 9:30pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/11 "2025-09-23T21:30:19Z")
…and allow us to filter against it, too, please. ---
**Author:** ![optilete](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/optilete/32/1908_2.png) [@optilete](https://forum.inaturalist.org/u/optilete)\ **Post date:** [September 23, 2025, 11:07pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/12 "2025-09-23T23:07:03Z")
> [@malcolmstorey](#): > > I’m working on UK Ichneumonidae. Most photos on the web are wrong. There are only a handful of spp in books. Most records in iNat are wrong. The only hope a beginner has is using ObsIdentfy or [https://waarneming.nl/](https://waarneming.nl/) but it’s a lot of work. I did not understand this line. Is ObsIdentify more reliable for UK Ichneumonidae, or does this app always return and Id on genus level for UK Ichneumonidae? ---
**Author:** ![sbushes](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/sbushes/32/8217_2.png) [@sbushes](https://forum.inaturalist.org/u/sbushes)\ **Post date:** [September 23, 2025, 11:22pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/13 "2025-09-23T23:22:36Z")
Great to see you’re helping fix the Ichneumon records @malcolmstorey ! > [@malcolmstorey](#): > > Most records in iNat are wrong. Do you mean research grade observations or… just those with a single ID? Looking at your corrections, I see many are for observations with a single ID using autosuggest - signalled by the little shield symbol on the ID : ![Screenshot 2025-09-23 at 23.19.09](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/b/9/b969f093bc1307a608f9ef2a94e000c680b29a81.jpeg) These are the AI suggestions, so take it with a pinch of salt as to whether the author actually thought species correct - most likely didn’t even look to see if it was similar. At present the only way to get the AI to make better suggestions to new users is to increase the data quality - by correcting the IDs and feeding it examples of confusion species so it knows not to be over optimistic in suggesting a species level ID. If you feed into the system in this regard, you will see it does pay off! When I started, UK diptera autosuggestions were really awful, but now ( particularly due to the huge input of Ian Andrews - @ophrys … ) I think it’s pretty impressive for the most part. Similarly, I imagine when Steve Gregory @bmig\_steve - started fixing the UK isopods it must have been an uphill battle, but now the data-set and hence the autosuggestions, are relatively reliable. Also, the better the dataset, the more other specialists seem to start to join in. With European sawflies, the shift has been really notable in that regard - since Andreas Taeger @symphyta started helping out, there have been various others joining in it seems. So, don’t despair too much, and if there’s anything us other UK users can do to help, please feel free to tag us in - people like myself and @matthewvosper and @josscarr will most likely be delighted to assist wherever we can 🙂 ---
**Author:** ![malcolmstorey](https://avatars.discourse-cdn.com/v4/letter/m/b5e925/32.png) [@malcolmstorey](https://forum.inaturalist.org/u/malcolmstorey)\ **Post date:** [September 24, 2025, 12:12am UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/14 "2025-09-24T00:12:09Z")
ObsIdentify was trained on [https://waarneming.nl/](https://waarneming.nl/). The Ichneumonidae has been cleaned up to a very high level and they have far more European records than iNat has British (I dunno - does iNat’s AI stick to a single country?) Anyway, they have more and better training data so of course it works better. It still bails out quite often and just says “unidentifiable Ichneumonidae” or similar. Yes, it goes to species when it can. ---
**Author:** ![sbushes](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/sbushes/32/8217_2.png) [@sbushes](https://forum.inaturalist.org/u/sbushes)\ **Post date:** [September 24, 2025, 12:16am UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/15 "2025-09-24T00:16:46Z")
Also, @matthewvosper mentioned the section which displays the “community taxon” - this is the bit that matters in terms of the datapoint which gets passed to GBIF or iRecord. ![Screenshot 2025-09-24 at 00.13.52](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/7/1/715c36ac8fdea52f7ab8886a548f0524a78350ba.jpeg) See [this help page link](https://help.inaturalist.org/en/support/solutions/articles/151000173076-what-are-the-community-taxon-and-the-observation-taxon-) to learn more about the difference between community taxon and observation taxon. ( usually the same… but in this instance different due to the user opting out ) But note also the “what’s this” button to the top right : ![Screenshot 2025-09-24 at 00.25.27](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/1/e/1e17e44a8d77c8d319ea5faf960d52eee8378787.jpeg) If you click on that, you will see a description and breakdown of how the algorithm works : ![Screenshot 2025-09-24 at 00.16.18](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/1/a/1a465cb61a275c0032f9698a05e85be513517f6c.jpeg) However. This also looks like a bug to me as it happens : The disagreement count should be 2 at subspecies level also…?? ![Screenshot 2025-09-24 at 00.21.38](https://us1.discourse-cdn.com/flex016/uploads/inaturalist/original/3X/5/5/55fbc1aedf9390db278b9d9fc927d2e0dfcdd225.jpeg) Anyone seen that before? @pisum ---
**Author:** ![malcolmstorey](https://avatars.discourse-cdn.com/v4/letter/m/b5e925/32.png) [@malcolmstorey](https://forum.inaturalist.org/u/malcolmstorey)\ **Post date:** [September 24, 2025, 12:31am UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/16 "2025-09-24T00:31:18Z")
I was trying to get record counts that could be used to assess frequencies (commonness) for the field guide that Lennart Bendixen and I are writing. So I went thru each species in the book checking boith iNat and iRecord (and geting quite confused which features were in each program!). I didn’t worry that much if there were 100’s of records cos those were the common ones that we knew were common. Not sure I worried whether single Id or RG, but most were single id anyway. The books only have a few species so everything gets shoe-horned into those species (I guess) and presumably the AI then follows the crowd. As an example: I went thru Therion circumflexum. Therion circumflexum has a black tip to the metasoma so I moved all those without up to the subfamily. There were 90 - now 28! Ichneumon is even worse, I guess cos lots of people think it’s the English name for all Ichneumonidae! So now my corrections are all floating around at some higher level taxon which is the parent of the original id and my correction. I wonder if I’ll ever find them again? ---
**Author:** ![sbushes](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/sbushes/32/8217_2.png) [@sbushes](https://forum.inaturalist.org/u/sbushes)\ **Post date:** [September 24, 2025, 12:44am UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/17 "2025-09-24T00:44:07Z")
> [@malcolmstorey](#): > > my corrections are all floating around at some higher level taxon which is the parent of the original id and my correction. I wonder if I’ll ever find them again? You need to use some URL magic to do this, but it is possible : `https://www.inaturalist.org/observations?ident_user_id=malcolmstorey&ident_taxon_id=685632` For example, this URL gives you all IDs by you of the genus _Virgichneumon_. If you want to search for IDs you’ve made of a different taxon to _Virgichneumon_, you just need to append it with the relevant taxon ID number. To do so, just go to the relevant taxon page and get the ID from the URL - e.g. for _Pimpla rufipes_ `https://www.inaturalist.org/taxa/496161-Pimpla-rufipes` Then change out the number at the end of the first link like so : `https://www.inaturalist.org/observations?ident_user_id=malcolmstorey&ident_taxon_id=496161` Actually that’s not quite correct though, as it gives you all obs with IDs of _Pimpla rufipes_…not just obs with IDs of _P.rufipes_ by you - maybe someone else knows how to fix - @bouteloua ? ---
**Author:** ![DianaStuder](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/dianastuder/32/43358_2.png) [@DianaStuder](https://forum.inaturalist.org/u/DianaStuder)\ **Post date:** [September 24, 2025, 5:41am UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/18 "2025-09-24T05:41:57Z")
> [@sbushes](#): > > These are the AI suggestions, so take it with a pinch of salt as to whether the author actually thought species correct - most likely didn’t even look to see if it was similar. ? lots of us skip _I can typing_ and use CV as an efficient way to get to the ID we know. Your disagreement bug query? The last column is Ancestor Disagreement [https://www.inaturalist.org/posts/25514-clarifying-ancestor-disagreements](https://www.inaturalist.org/posts/25514-clarifying-ancestor-disagreements) iNat’s own explanation. With 5 years of comments. ---
**Author:** ![cthawley](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/cthawley/32/1445_2.png) [@cthawley](https://forum.inaturalist.org/u/cthawley)\ **Post date:** [September 24, 2025, 12:10pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/19 "2025-09-24T12:10:47Z")
This > [@sbushes](#): > > “community taxon” - this is the bit that matters in terms of the datapoint which gets passed to GBIF or iRecord. is only partially true. While agreement on the status of the CID is what determines whether an observation is exported or not (ie, is RG), the Observation ID, not the CID, is what is exported, see: [https://forum.inaturalist.org/t/which-id-of-an-observation-is-exported-to-gbif/18563](https://forum.inaturalist.org/t/which-id-of-an-observation-is-exported-to-gbif/18563) ---
**Author:** ![jennifervb](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/jennifervb/32/2863_2.png) [@jennifervb](https://forum.inaturalist.org/u/jennifervb)\ **Post date:** [September 24, 2025, 2:52pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/20 "2025-09-24T14:52:46Z")
This might be off topic, but what is the purpose behind opting out of the community taxon? I didn’t know this was a thing and I don’t understand the motivation. ---
**Author:** ![cthawley](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/cthawley/32/1445_2.png) [@cthawley](https://forum.inaturalist.org/u/cthawley)\ **Post date:** [September 24, 2025, 3:21pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/21 "2025-09-24T15:21:36Z")
There are a number of existing threads on the forum that might be helpful here: [https://forum.inaturalist.org/t/why-does-community-id-honor-opt-out-of-community-id/23649](https://forum.inaturalist.org/t/why-does-community-id-honor-opt-out-of-community-id/23649) [https://forum.inaturalist.org/t/opting-out-of-community-id-vs-literally-anything-else/26941](https://forum.inaturalist.org/t/opting-out-of-community-id-vs-literally-anything-else/26941) [https://forum.inaturalist.org/t/why-opt-out-of-community-id/35116](https://forum.inaturalist.org/t/why-opt-out-of-community-id/35116) [https://forum.inaturalist.org/t/opting-out-of-community-id/3078](https://forum.inaturalist.org/t/opting-out-of-community-id/3078) If a more in-depth discussion of motivations for opting out of CID is needed, a new thread could be created or one of these re-opened. It’s best to keep the thread here focused on the OP’s original topic though. ---
**Author:** ![sbushes](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/sbushes/32/8217_2.png) [@sbushes](https://forum.inaturalist.org/u/sbushes)\ **Post date:** [September 24, 2025, 3:34pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/22 "2025-09-24T15:34:55Z")
> [@cthawley](#): > > While agreement on the status of the CID is what determines whether an observation is exported or not (ie, is RG), the Observation ID, not the CID, is what is exported Interesting, my bad. > [@DianaStuder](#): > > Your disagreement bug query? The last column is Ancestor Disagreement My query is in regard to the 3rd column, not the last. ---
**Author:** ![optilete](https://sea2.discourse-cdn.com/flex016/user_avatar/forum.inaturalist.org/optilete/32/1908_2.png) [@optilete](https://forum.inaturalist.org/u/optilete)\ **Post date:** [September 24, 2025, 10:21pm UTC](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608/23 "2025-09-24T22:21:42Z")
It’s more complicated, but not the topic of this thread.. [https://forum-archief.waarneming.nl/index.php/topic,499061.0.html](https://forum-archief.waarneming.nl/index.php/topic,499061.0.html) [https://forum.inaturalist.org/t/do-you-use-inat-or-a-local-equivalent-while-traveling-abroad/41429/8?u=optilete](https://forum.inaturalist.org/t/do-you-use-inat-or-a-local-equivalent-while-traveling-abroad/41429/8) [Next page](https://forum.inaturalist.org/t/community-identification-model-doesnt-work/70608.md?page=2)