Changes to the Agree button/functionality and addition of Markdown in comments and ID Remarks

[joe_fish]

I just want to add to the chorus of “this is a terrible change to the functionality of iNaturalist. please reverse it immediately”.

I’ve identified 30,000 or so observations on here. I curate several large taxa, and much of that curation involves clicking Agree to quickly confirm the ID. I’ve gone through the back catalog of thousands of observations for these groups. Having to manually type the names in GREATLY slows down my productivity on here and dissuades me from wanting to expand my efforts. I would simply not bother to go through and do this work if this remains the case, which ultimately prevents me from weeding out the misidentifications that lurk on here.

For the comments here questioning why an observation needs more than 2 IDs: imagine a 3rd ID is later added that is incorrect. That observation now loses its relevance. There is always a benefit to having multiple confirmations, particularly if those IDs come from trusted “Experts” in the taxon.

The idiom of cutting off one’s nose to spite their face comes to mind here. Whatever harm there is with “competitive identifiers” that this change is attempting to address is being far outweighed by the large inconvenience for users like myself.

[upupa-epops]

@joe_fish

For the comments here questioning why an observation needs more than 2 IDs: imagine a 3rd ID is later added that is incorrect. That observation now loses its relevance.

That observation would now become Needs ID and show up in identify the feed of more identifiers, most of whom sort by Needs ID as far as I know?

[maractwin]

I’ve found another problem with this change.

For the time being, I have quit reviewing all observations in the taxa that I used to follow daily looking for data consistency, and switched to a secondary interest where I have only ever tried to just provide initial IDs. This is North American Butterflies, where there are many identifiers, and a huge number of observations.

Because there are a lot of identifiers, I often find that before I get to the bottom of the 30 observations shown on the identify page, several of them will already be identified or confirmed by someone else. When I get to those observations, there’s no Agree button on the page. I don’t have a problem with that, except that I then see 3 or 4 in a row, and wonder what’s the point of stepping through these, spending a little time to decide what’s in the observation, and then discover that I can’t record it. This is frustrating and feels like I am wasting my time.

Can you change the identify page to dynamically remove observations as they become research grade? I know that is difficult from a programming standpoint, but if you’re not going to allow me to identify it, why show it to me?

[maractwin]

@upupa-epops

@joe_fish

For the comments here questioning why an observation needs more than 2 IDs: imagine a 3rd ID is later added that is incorrect. That observation now loses its relevance.

That observation would now become Needs ID and show up in identify the feed of more identifiers, most of whom sort by Needs ID as far as I know?

The problem is that it’s an old observation which is now Needs ID, and is lost way down in the list of thousands of observations. If I was able to easily subscribe to it, I would be notified that someone has introduced controversy and that I need to look at it a second time. As the system is now, I no longer look at observations with just 2 IDs and will never know if their status changes.

[upupa-epops]

I do feel like there’s no reason not to have an Agree button on all of the identifications on this observation: https://www.inaturalist.org/observations/53524793
I get the impression that that result is one of the less intentional changes though.

[upupa-epops]

Fair. I wonder which is a bigger priority though, dealing with these conflicting IDs (where the community ID should still be correct, just not species-level precise), or getting to those thousands of Needs ID observations. I see the value in reviewing both and get the dilemma.

@maractwin

Because there are a lot of identifiers, I often find that before I get to the bottom of the 30 observations shown on the identify page, several of them will already be identified or confirmed by someone else.

I used to have this issue a lot when I identified Unknown observations. I dealt with it by skipping to like the 3rd or 5th pages of observations, or by sorting the observations in ascending or random order rather than the default descending order. Basically just changing the filters up so you’re not using exactly the same workflow as the other identifiers.

[po-po-pro]

Thanks; I’d somehow overlooked that!

[fogartyf]

But it will not show up for anyone who has marked it as reviewed. If you are dealing with a taxon that has few expert identifiers on iNat, it’s possible that the experts will all miss this because they have marked it as reviewed and do not receive a notification of the change.

[bobmcd]

@maractwin

When I get to those observations, there’s no Agree button on the page. I don’t have a problem with that, except that I then see 3 or 4 in a row, and wonder what’s the point of stepping through these, spending a little time to decide what’s in the observation, and then discover that I can’t record it. This is frustrating and feels like I am wasting my time.

Probably not helpful because there are a couple of more steps but, on the identify modal one can select to view the observation page and then select agree with the community taxon.

[bobmcd]

@fogartyf

it’s possible that the experts will all miss this because they have marked it as reviewed and do not receive a notification of the change

Again, maybe not helpful but one can follow an individual observation to get notification of change.

[fogartyf]

The problem is that following will give a notification for any identifications, even concurring “agrees”, regardless of your notification settings. So that is not a workable solution.

EDIT: to clarify, it’s not workable because I would receive hundreds of notifications for concurring agrees on RG observations, which would make it infeasible for me to use the notification feature.

[rupertclayton]

As someone who does not care for gamification or leaderboards, I want to express my concern that this change will have major negative impacts for that minority of iNat users who actively work to ID observations.

iNat staff have already acknowledged that engaging users to assist with IDs is an ongoing challenge. While I understand the motivation behind the change, I fear that an attempt to address the relatively minor issue of people piling on IDs will cause many productive identifiers to scale back their contributions to iNat.

iNat identifiers have a range of interests and focuses. In general, though, most people choose to identify organisms within a particular taxon and geographic area. As people build specialist knowledge (or bring it from their prior study), a significant factor for continued engagement is the ability to “curate” a particular taxonomic area. Users can apply their knowledge to improving the quality of identifications across a species group, genus or family and see those results in the form of more precise distribution data, more accurate computer vision training, better educated fellow iNat users and so on.

One technique that identifiers use to build this quality is to review existing observations and provide confirming or diverging IDs using their own experience, field guides, dichotomous keys, etc. In iNat this might be accomplished in the Identify tool by selecting a parent taxon (and maybe a geography) and including all observations (Needs ID, RG and Casual). Until now, the identifier could then navigate through these observations, choosing to Agree or add a new Identification as appropriate. The end result of this is that the identifier has made an identification for every observation of that taxon that they can confidently ID.

As an identifier, the new agree functionality prevents me from using this workflow. I can no longer agree to the ID for any observation that is already at RG status. All I can do is ignore it, mark it Reviewed, or manually add a confirming identification. It might appear that checking Reviewed achieves much the same result as clicking Agree, but that falls short in several ways:

• When I look at some observation later, I no longer get an active confirmation that I concurred with the ID. I can’t tell whether my Reviewed checkbox means “I agree with this ID” or “I really don’t know but I want to stop seeing this” or “I pressed the ‘R’ key by mistake”.

• Other iNat users don’t know whether a particular identifier has reviewed an RG observation or not. Consider insects or spiders, taxons that are notoriously difficult to accurately ID to species level. I wouldn’t put much credence in the identity of an RG observation whose ID is based on the observer’s initial computer vision ID plus a confirming ID from a new iNat user who was part of the same bioblitz. If those two IDs are supported by an ID from a knowledgeable expert in the taxon, my confidence would be much higher. The new functionality means I cannot tell whether the ID has been reviewed and explicitly confirmed by an experienced identifier or not reviewed at all.

• I can rely less on any searches or reporting I might do looking for my own username or someone else’s as the identifier. In the past, I could assume that a search for observations identified by a particular user would tell me which ones that person had reviewed. Now, the search will exclude an unknown number of additional observations that they reviewed but didn’t ID because they were already at RG status.

I’m aware that I can still manually add an identification that confirms the current ID. Realistically, I can’t see myself doing that very often. I’ve reviewed 20,000 plant observations in California and western states over the past year because the Identify tool gave me a really efficient way to apply and build my knowledge and see the results. I can tell that I have concurred with or dissented from the ID for every observation of a particular taxon. For the next 20,000 observations the only way I can continue to do that is to type in a species name 18,000+ times. That’s quite a disincentive.

[pisum]

@maractwin

For the time being, I have quit reviewing all observations in the taxa that I used to follow daily looking for data consistency, and switched to a secondary interest where I have only ever tried to just provide initial IDs. This is North American Butterflies, where there are many identifiers, and a huge number of observations.

isn’t this actually a good behavioral change? fixing other people’s bad IDs is definitely a helpful thing, but maybe that could be shifted to be more like a monthly activity, and then the primary daily activity would be to add new IDs.

(i would add, just as @astra_the_dragon noted above, and I’ve also said before, that it might be even more helpful if really knowledgeable experts hold off on their IDs of really easy-to-ID taxa for some period to let others get first dibs on those. if i’m a beginning identifier, it helps to have a nice supply of observations that match my skill level. it’s discouraging if all the ones that are left are the ones even the experts have trouble with.)

[tiwane]

Thanks for your comments everyone, especially if you tried it out for a while and provided feedback, or called out specific use cases which this change negatively affected. We plan on rolling in back later today (California time), as Ken-ichi noted here.

I’ll leave this discussion open for a bit longer, but if it becomes a rehash of ideas and thoughts that have already been expressed, I’ll close it.

[calebcam]

Thank you so much for listening to us and I am very pleased with Ken-ichi’s decision to revert the change.

[cooperj]

I’ll add my thanks for listening.

I was one of those attempting to curate a fairly high-level taxon at national level because it is my area of research. Agreeing with a previous determination provided the benefit that if somebody disagreed with my determination of X I would be notified. Then I would go back to those records and add more detail about why I agreed with the determination of X but don’t agree with the newly added determination Y. I can’t do that for every identification. Not being aware of these ‘Y’ records can result in correct records being lost as RG, or worse, incorrectly identified records becoming RG. The latter usually because of some systematic misunderstanding. It is those systematic misidentifications that lead people to suggest Y rather than X, and then other people agree with them, for the same reason. So, overall quality will decrease, at least from my perspective. Like others I’m not saying the change is a bad thing. Blind agreement by some users is a big problem. But it would have changed the way I interact with iNat, and my use of the data for research.

[choess]

Tony,

Let me just say that I appreciate both:

1. iNat’s willingness to experiment with potentially controversial changes and
2. openness to feedback and reversion on said changes.

This is always hard in places with big, established user communities, and I’ve seen it handled much worse.

[cthawley]

Super appreciate the willingness of staff to hear users’ concerns on this.

While I’m one of those that feel removing the agree button was a negative change, I also definitely respect users who feel that there are problems with gamification, and RG observations with incorrect ID’s supported by a bunch of incorrect agrees are also an obvious problem. There’s obviously community support for addressing this, and, based on all the comments and forum activity, many people have been thinking a LOT about this over the weekend. I am wondering what the best way is to take all of the ideas and thoughts and maybe coalesce them around some changes supported by the community that could address these issues? To sound hippyish (in a good way), is there a way to take all this energy (some negative), and spin it into positive engagement?

As an example, I was wondering if it might be appropriate to make a wiki post of potential changes that could address some of these issues, like a community brainstorm sesh? I have seen so many ideas floated in comments that I can’t keep track, and a bunch of feature requests seems like it would be confusing. Maybe a post like that could be seen by staff who have a much better idea of how feasible/implementable changes might be. Are there any low-hanging fruit we are missing?

A few examples rattling around in my brain:
I’ve heard folks complain about getting too many agreeing notifications (I confess I don’t understand why turning these off which is already possible isn’t a good solution, but…).
Potential solution: Offering more fine-grained control over notifications (like a setting that says, only notify me of identifications that change an observation’s RG status in either direction).

@bouteloua also mentioned that new users sometimes feel overwhelmed with agree notifications (and unlike “power-users” may not know where to find notification controls).
Potential solution: Change the account defaults so users don’t get agree notifications (or combine with defaulting to the option above, so they only get notified on agrees that affect their observations’ statuses).

Or to address one of the biggest agree problems (the reflexive Agree of the original observer to another user’s ID suggestion): remove the Agree button for the original observer only! (not sure how much god-like power staff has over the Agree button).

Maybe some or all of these have already been put forward, considered, and/or rejected, but I’m sure other users have lots of creative, targeted ideas on this topic that might be useful to have “out there”.

Anyways, thanks again for listening!

[arboretum_amy]

@cthawley

Offering more fine-grained control over notifications

This is coming, see Tony’s post here

[sbushes]

@cthawley

Or to address one of the biggest agree problems (the reflexive Agree of the original observer to another user’s ID suggestion): remove the Agree button for the original observer only! (not sure how much god-like power staff has over the Agree button).

I think this is really problematic, and it concerns me when people mention it.
For me for example, at times I have actively posted at genus, with an idea of species… but wanted to wait for an expert to chip in before taking it that far. If I went straight to species, my concern would be that one of the non-experts would confirm and it would be taken out of the Needs ID pool without real review.

As an observer, I think this just straitjackets the OP in a weird way…
Its also fine maybe for birds, but terrible for less attended taxa and locations.

Why is nobody considering simply making the withdraw button more visible instead?
(I suggested a connected feature request before I read this… )
Isn’t this issue all partly a question of OPs simply not knowing what to do?

But I agree, an assimilated wiki and connected conversation shifting this towards positive change would be lovely to see.