Community ID (and that is a ‘community’ of 2 which is pushing the definition of the word hard)
1 Like
I agree. “Research Grade” is totally the wrong phrase. Makes me cringe.
4 Likes
I think it would be totally worth it, and something we should do. I just was trying to illustrate that sometimes simple-seeming changes require a lot of steps.
2 Likes
More worrying to me is the fact that such low-quality RG observations get exported to third-party databases… which (being hosted by some highly-regarded institution) may then be relied upon to assert the regional occurrence of taxa.
It may even lead to a self-reinforcing trust in a blatantly wrong ID: “You say my observation on iNaturalist is wrong because the species does not occur in this area, however I found this other Museum database claiming the species is found in the area!” Then, upon investigating the other database, you come to realize it contains… only that one misidentified observation automatically exported from iNaturalist due to an undeserved RG status.
2 Likes
Each scientific collection has wrong ids, as it was discussed, it’s an open database and each corrected id will be shown on other websites too, problematic taxa are known, all that is needed are people checking them and correcting ids, also teaching other users what to do and not to do.
3 Likes
… all that is needed is hundreds of volunteers trying repeatedly to correct what could have been, at least partly, prevented by technical measures - in the form of either higher requirements (number of concurring IDs), or an educational message box dissuading from blindly elevating observations to RG status. Yeah, got it, it’s hard to implement :)
And I definitely agree that each scientific database contains errors and is not to be trusted - quality requirements for inclusion must be low to keep data coming in. Bad data is better than no data.
2 Likes
but we start from - iNat’s stated first intention is to encourage all those NON-scientists to notice nature.
2 Likes
Not necessarily.
At the moment in the UK, we have national recording scheme experts checking and changing bad IDs before they reach our national database NBN. There is no feedback system though, so NBN IDs will be different to iNaturalist IDs. Both NBN and iNaturalist pass to GBIF independently as I understand, so either there will be duplicates or different IDs for the same datapoint.
At the same time, before the data bridge was reinstated here between iNat and iRecord, at least one LERC ( local environmental records centre ) I spoke to also downloaded iNat data and used for their regional database anything that went to RG. I tried to explain RG data was in flux, but they just said there was no way of keeping up with changes for them so they had to use a single import. These records can also go to NBN and in turn GBIF it seems.
In addition, before the data bridge was reinstated some of us were also forced to upload to multiple platforms, creating duplicate datapoints ourselves.
In theory, this could mean you could have multiple datapoints even on GBIF for the same observation with both the original iNat ID (via LERC), the updated iNat ID (via iNat) and a third ID (via iRecord).
Its unclear what checks are in place to prevent this.
UK dataflow isn’t streamlined or centralised enough, so
- most datapoints offsite will not be updated when the iNaturalist ID itself is corrected
- not all IDs will remain static off-site
- most ID corrections off-site will not be passed back to iNat
Here’s a nice diagram from NBN showing how complex UK dataflow is !
3 Likes
Sorry, I forgot about it, I don’t get why connection between NBN and iNat wasn’t improved the way it should be long ago, and I kinda antagonize those experts a little bit, they don’t wanna learn another platform because of “too many” observations and how they were arguing because of obscuration…
2 Likes
If anyone is using any database to blindly assert occurrence of taxa, they’re pretty ignorant of the realities of these databases. You should see the errors within scientific museum collections which are being propagated to databases like GBIF. And to identify and correct those errors, you have to fly in a plane to the museum, find the specimen, and convince the curator that they need to update their identification and in-house database. All databases are like Wikipedia–a fine place to start, but then the work begins.
Back in the day when people wrote books such as “Mammals of Colorado” or “Mammals of New Mexico”, the author would travel to each and every museum, physically examine each and every specimen to confirm ID, and then put the dot on the map documenting its occurrence. People still create these maps and write these sorts of books, but the time and level of dedication that the authors have nowadays seems to have diminished considerably. Without verification, the data will contain errors regardless of whether it came from iNat or the Smithsonian.
But its always good to identify ways to reduce the errors.
6 Likes
Like it or not, it happens already: laymen not thinking of questioning the sources for a database, being overconfident in the quality of data, some even going as far as reusing it professionally (ugh!). I’m all for educating data users not to trust data blindly, to assess its reliability etc. Is it reason enough not to try and prevent -by technical means- the inclusion and subsequent dissemination of unreliable data in the first place?
edit: last message in this thread. As said, I’m beating a dead horse. iNat is a wonderful source of renewed interest in nature - for the sake of it, let’s cast a veil over minor side effects and shortcomings.
1 Like
@pfau_tarleton – I agree with your post. Research museums are certainly not free of errors and those errors in ID can be perpetuated through references based on specimens. Even when the authors of field guides have personally checked IDs on specimens before constructing range maps, errors can happen (I know of some in the two mammal books you reference). Nothing’s perfect and we make mistakes. A few years ago I started investigating some questionable records of dragonflies and damselflies in NM by reaching out to curators of collections where the voucher specimens reside. Turns out a few dubious ones were actually correct but quite a few others were misIDed in the past and the errors were never fixed until I inquired. The misIDed specimen records had been included for years in other secondary references, skewing some range maps quite badly. The lesson of the story is it often can take years, a little detective work, and some concerted effort to undo bad IDs once they are made.
4 Likes
Agree with every point that @whaichi made.
A minor recommendation: make sure to @ whomever you’re commenting to in an observation since some identifiers unsubscribe from the observation after making the ID, so they don’t get notified about comments or other identifications after making theirs.
Sometimes there won’t be a response to @ or questions in comments though, sometimes people are very busy or maybe the request gets buried at the bottom of a long pile.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.