Explicit Disagreement Update

Turns out the way we handled explicit disagreements wasn't quite meeting one of the goals we had, which was to allow people to add more conservative identifications without changing the taxon the observation was associated with, so after running things by the Google Group we made a somewhat more radical change. The Community Taxon will continue to operate as it did, but the taxon an observation is associated with should now be the most specific proposed taxon with which no one disagrees. Here are some examples:

ID 1: Homo sapiens
ID 2: Genus Homo, not a disagreement
=> CID: Genus Homo (this is what the community of 2 agree on)
=> Observation Taxon: Homo sapiens (the finest proposed taxon no one disagrees with)

So now this observation will show up as Homo sapiens in search results and such. It's still in "Needs ID" b/c there's no consensus at the species level, but the user who made ID 2 can chime in without bumping the observation back to species-level.

Here's the slightly more controversial scenario:

ID 1: Genus Homo
ID 2: Genus Homo
ID 3: Homo sapiens
=> CID: Genus Homo (this is what the community of 3 agree on)
=> Observation Taxon: Homo sapiens (the finest proposed taxon no one disagrees with)

This is a bit less conservative than we've been in the past because observations can move toward a finer taxon with less community consensus. Some people (cough*Scott*cough) like this because hey, no one's said it isn't Homo sapiens so why not go with that until there's a contradiction? Others (cough*me*cough) are probably going to feel like this just makes it easier for incautious identifiers to shift more observations toward incorrect taxa. To my fellow naysayers, this might also mean that fewer people click "Agree?" just to move the observation to species level, so maybe we'll end up with more incorrectly-identified observations in "Needs ID" but fewer in "Research Grade," which might be a good thing.

Anyway, try it out over the next few days and see how it works and feels. For a more extensive background on these changes, including illustrated examples, see https://docs.google.com/document/d/1BHX1pnoMEv9gf_PqDzoqdQmZjsvdU_j8O7TEGwdr6fg/edit

Posted by kueda kueda, January 12, 2018 23:41


I found a problem with this right away, where a research grade observation lost its research grade because of a subspecies ID. This observation of a green frog (https://www.inaturalist.org/observations/9381880) had 5 IDs at species level, making it research grade. It then was IDed a single time to subspecies level. This means it is no longer considered a research grade green frog, despite the fact that it most certainly is. More precise IDs shouldn't be able to remove research grade.

Posted by mws almost 4 years ago (Flag)

I’m not a huge ‘subspecies’ kinda guy, so it will be kinda weird when folks add on a subspecies ID and the observation still ‘needs an agreeing ID.’

So, ID 1: Homo sapiens
ID 2: Homo sapiens
ID 3: Homo sapiens
ID 4: Homo sapiens subspecies sapiens

And the observation ‘needs ID.’

Again, it’s that silly ‘final reward’ of getting something to ‘research grade,’ but there are also some pretty substantial disagreements when it comes to subspecies delineations.

Overall, I’m fine with it, and will likely become used to it, but I do hear @jason_m ‘s concerns and may get similar concerns from others.

Posted by sambiology almost 4 years ago (Flag)

I wonder if the admins could make it so that the observation has to get research grade twice? Once for species level, and then again for subspecies level. That way you can only get more accurate. But then again having two grades on one observation might just create clutter.

Posted by mws almost 4 years ago (Flag)

So far this seems like a great change. If nothing else, this makes it a lot easier to search for observations to identify. Also I waste far less time withdrawing IDs.

Posted by glmory almost 4 years ago (Flag)

This is good, and especially for cleaning up all the old observations stuck at a high level observation taxon. But is there a way to do all these old ones automatically? For example, I had suggested Parry's Primrose some time ago on this observation https://www.inaturalist.org/observations/321521, but as of a few minutes ago it was still at "Dicots" for the observation taxon. To get it to change the observation taxon, I had to delete my ID and then re-enter it. If there's a way to make the observation taxon change on these types of old observations without us having to delete and re-enter IDs (which I doubt anyone wants to do), that would immediately make many, many observations more searchable.

Posted by paloma almost 4 years ago (Flag)

I like it! Makes people aware of ssp IDs.

Posted by calebcam almost 4 years ago (Flag)

Responding to paloma's question, I don't know if there's a way to do it automatically, but you don't need to delete your earlier ID. You can just reenter it or make any edit to your current suggestion. I went back through five months/600 of my suggestions last night (awake til 1:30 a.m. - 4 hours past my bedtime ;-) They were mostly still sitting at Hymenoptera level, so I clicked edit, added a dot (or comment or link) and saved, and this updated the CID. It was a one-time only hassle and has already helped an expert find and refine some IDs.
Other note about deleting IDs: I noticed that you then won't get follow-ups even when you've re-entered a suggestion. You need to select "follow" at top right of observ.
I love the change now - Seems much more productive less frustrating - Yeah!

Posted by susanna_h almost 4 years ago (Flag)

I am starting to like this new system, but the way it manages subspecies still irks me. I think an observation should be able to be research grade at species level but needs ID grade at subspecies level.

Posted by mws almost 4 years ago (Flag)

Interesting idea. If the IDs are

Homo sapiens
Homo sapiens
Homo sapiens
Homo sapiens
Homo sapiens sapiens
--- then perhaps it should be still shared with GBIF, but classified in the Needs ID section of the ID page? RG for species on GBIF, but needs ID for ssp on the iNat ID page?

Posted by calebcam almost 4 years ago (Flag)

I was thinking more of splitting the ID into two parts, species-and-above-level ID and subspecies-level ID. If it gets Research Grade in species-and-above-level, then it would be counted as a valid data point. if it gets research grade subspecies-level ID, then it just gets better, but if it doesn't research grade subspecies-level ID, then it isn't invalidated as a data point.

Posted by mws almost 4 years ago (Flag)

This change actually helps a lot with adding identifications to unknown species. Previously people often got annoyed if you tagged something at the kingdom level. This is partly because if they caught their mistake and added an ID it was still listed at kingdom rather than species level.

Now if I ID something as a plant and the submitter puts in a species name it will be named as that species. This means there is no longer a downside to me adding a high level ID.

Posted by glmory almost 4 years ago (Flag)

Another great example for how this makes it a lot easier to bring things up from being unknown is this observation:
Initially it was tagged as life.

Someone tagged it later as the genus Thuiaria. It still showed up as being life.

Previously I had no way to knock things up, I have no idea if the genus is correct so I couldn't agree with it. Adding a high level ID would be saying I disagree with the genus ID.

Now however I am able to add an ID of Hydrozoans. Immediately that moved the ID to genus Thuiaria, which is probably correct but I cannot confirm.

Posted by glmory almost 4 years ago (Flag)

Hm, I thought I'd left a comment here earlier, but I guess not. Some updates!

1) We made a special exception for subspecies: a single subspecies ID can no longer shift the Observation Taxon, so if the sequence is Homo sapiens, Homo sapiens, Homo sapiens ssp. inaturalensis, the observation will remain associated with Homo sapiens. For higher levels there is no change, so something like Genus Homo, Genus Homo, Homo sapiens, will associate the observation with Homo sapiens. This should make it "safe" to suggest subspecies to Research Grade observation without fear of changing the quality grade.

2) Disagreements should get updated after taxa move around the tree, you say it's a dinosaur, I disagree and say it's a bird, then birds get nested within dinosaurs, and my identifications is no longer considered a disagreement.

3) Allow obs that have opted out of the Community Taxon to be Research Grade if the owner's identification is a species and the Community Taxon is a subspecies of that species (sort of related to all this)

4) Fixed a bug where disagreeing with your own ID wasn't recording what you were disagreeing with.

Posted by kueda almost 4 years ago (Flag)

As usual, beautiful updates, Ken-ichi. Great stuff.

Posted by sambiology almost 4 years ago (Flag)

Thanks for these updates! They'll be helpful.

Posted by mws almost 4 years ago (Flag)

Add a Comment

Sign In or Sign Up to add comments