When location is altered, DQA "Location is accurate" should reset and notify downvoters

Feature Requests
1

Platform(s), such as mobile, website, API, other: Website

URLs (aka web addresses) of any pages, if relevant: Any observation page

Description of need:

At present, when one downvotes “Location is accurate” in the DQA, and the observer subsequently updates the location, the vote stays. The observer sometimes does not notify anyone of the update to the location, so their correction may not “fix” the observation and remove it from casual.

Feature request details:

Part 1. My suggestion is that when the location of an observation is edited, that should reset any votes on “location is accurate”.

Part 2. At the same time, it would be very useful, but not essential to this request, if any downvoters could be notified of the location update so they can check their concerns were addressed. I realise we don’t have DQA notifications yet, but understand they are planned.

2

I approved this request because the main request is related to the DQA. The ancillary

is something that

so please focus any discussion/cast your votes on the main part of the request:

Also noting that there is precedent for changing DQA votes based on user activity on the observation, such as this implemented Feature Request:

https://forum.inaturalist.org/t/remove-yes-votes-from-the-can-the-id-be-improved-dqa-when-the-community-taxon-improves/27273

My personal response to the request is that I would be a little concerned that this proposed feature would allow users a way to remove DQA votes that they disagreed with. They could change the location to something that could also be inaccurate or change the location and then change it back to the original incorrect location to reset votes. As a curator, I’ve seen a reasonable amount of observations with intentionally incorrect IDs (basically trolling, jokes), and I’d be concerned that users will to do that in the first place might exploit this feature. As there is no location edit history for observations, there wouldn’t really be a way to detect this (unless staff have something, but that still doesn’t help curators moderating issues on iNat if staff would need to be called in to address stuff like this - not a great use of their time probably).

3

I would prefer to be notified. Given the notification issue, I also leave a comment about the location, which then does trigger notifications if there are new comments / IDs.

Location Inaccurate = Casual, so that is a prompt to investigate – why is this obs (still ? ) Casual ?

4

I agree with @DianaStuder that notification (without resetting the DQA) would be preferable. I also manually leave comments when I tick ‘No’ to this question in the DQA, which is a bit clunky.

5

This is what I do. A lot of the organisms I ID may rely on accurate location for proper ID, as well. When I leave the comment, I specifically tell the user that they can ping me when they update the location, at which point I’d be glad to ID the organism for them. This also lets other IDers stopping by know that the location is inaccurate, so they are less likely to just automatically ID it.

Especially with new users that may just have made an error (or anyone really, no one likes to be corrected), I hope that explicitly offering to come back might make it feel less like a reprimand and more like encouragement in using the site, and show that I care about the user’s contribution.

If it automatically resets, I also worry about people “correcting” the location to something inaccurate, both as a joke or troll, but also just because they’re very sure of their correctness. I think that it would be be statistically uncommon for someone to do that as a joke, so I’m not sure it’s a reason not to have the feature, but it’s something that would be quite difficult to correct if they could just keep doing it. This is especially a problem if the inaccurate location isn’t immediately obvious. I’ve put a lot of detective work into establishing a location as incorrect before.

6

But it would remove all the votes they agreed with as well. What really matters is whether a reset changes the current consensus: the specific number of votes for and against doesn’t have much relevance if the difference is only one vote. In that particular case (which must be quite common), the current consensus can easily be overturned anyway if the observer hasn’t already used their vote. And where the difference is greater, it can still be overturned by the observer inviting others to follow their lead (just as with identifications). So this new feature wouldn’t open up an opportunity that doesn’t exist already.

Our default assumption should always be that the observer is responding positively to requests and acting in good faith, and we would like to reward that by updating the DQA as promptly as possible. The main reason why a reset might be favourable is that it could sometimes make overturning the current consensus a lot easier, since the original voters may not all be available. After a reset, if the DQA was still needed, it would only need one person to reinstate it.

With any automatic DQA resets, though, I think notifications would be essential, rather than just highly desirable.

7

It isn’t just notifications, though. If the DQA is reset, there is no record that there were previous votes. This can cause a fair amount of confusion, particularly if the comments on the observation lead one to expect to see DQA votes, so it could be difficult to reconstruct whether the observation was in fact edited or the commenting person merely forgot to vote. I see this problem already with the “ID cannot be improved” DQA which was changed to be reset if the community ID changes.

(Need I reiterate that we sorely need a record of edit history on observations?)

8

Agreed. this would address a number of things.

9

Thanks all for the feedback. Based on this, what I’m taking away is that most of you see it as essential to have better notifications and/or observation edit history before implementing this feature request, to reduce misuse by bad faith observers. That seems fair. Even if bad faith observers are a small minority, they can cause disproportionate problems and frustration.

10

The problems I see have less to do with potential misuse by bad faith observers if implemented, and more to do with lack of communication and documentation when something changes (even if edits are done in good faith). Resetting the votes would erase relevant information (there was a problem; someone reacted to it by voting on the DQA, often in conjunction with a comment) without replacing it with anything that would close the communication loop (letting other people who have previously interacted with the observation or who visit it in the future know that the problem was fixed).

I can see resetting the votes if the edits have made the DQA no longer applicable (for example, it should be reset on the DQA for photos of different species if the observation has been edited so that only a single photo remains) but it is more difficult to justify for other cases unless it is accompanied by other changes to the way the system works – as has been pointed out, it is easy enough to counter a single vote (though observers do not always know enough to do so themselves, nor do they always leave a comment after editing even if they have been expressly asked to do so).

11

The feature request is intended to erase irrelevant information. If a substantive edit has been made to the location, the previous votes no longer apply. This leaves the observation in an invalid state with non-trivial consequences - in most cases, it will remain casual when it shouldn’t be (assuming the location is the only DQA issue).

It isn’t necessary to know the previous voting history to evaluate the current state of the observation. The current location is either accurate or it isn’t, with the default assumption being that it is. That’s how every observation starts out, so resetting the votes wouldn’t introduce a radically different state of affairs.

An editing history would only be useful in the relatively rare cases where no one ever responded to the DQA notifications (assuming we had them). But in that case, what would prompt anyone to look at that history - or indeed to (re-)review the observation at all?

This is the crucial issue that the feature seems intended to address. It just seems plain wrong that an observation remains casual after the observer has tried to do everything necessary to improve it. (I’m deliberately ignoring any hypothetical bad-faith actors, for the same reasons I gave earlier).

The desirability of editing histories is an entirely separate issue. In practice, they often turn out to be far less useful than most people anticipate, and only really come into their own on sites with dedicated review queues (which don’t depend so much on personal notifications).

12

Did they leave a comment, prompting a notification, which gives the other person a chance to reconsider their Inaccurate Location vote ?

13

That describes the current state of affairs. Comments don’t improve the observation itself. They’re only necessary as a work-around for the lack of notifications. Neither the observer nor the DQA voters receive notifications, so it requires both parties to leave comments (and respond promptly to them). Also, it may require more than one voter to overturn the current DQA state. So, all in all, an over-complicated and unnecessarily slow process that this feature request would certainly help to improve.

14

Is that the case, though? How is knowing that someone marked the DQA in the past irrelevant information just because the observer edited the observation?

Depending on what problem motivated the DQA vote and the observer’s competence with whatever interface they use to interact with iNat, there is no guarantee that the problem has been corrected after editing. I’ve seen plenty of cases where the observer has edited an observation in good faith in response to requests but has not ultimately resolved the problem, or the editing has affected other things that are relevant for ID and/or the status of the observation (e.g., questions about where a specimen was collected and whether the rest of the observation data reflect the place and time of collection or the place and time of photography and whether it should be treated as wild or captive).

While problems connected with location are usually more obvious than problems connected with other DQA items, it is nevertheless not always self-evident what motivated the DQA or whether the location after editing is correct. Erasing the DQA only adds an additional source of potential confusion without removing the existing ones.

15

Yes, it is. It’s much more confusing to have DQA votes that don’t relate to the current location, than to have no votes at all. Anyone viewing the observation for the first time would be very puzzled by any disagreeing DQA votes if the current location looks right and there’s no comments suggesting how that situation might have arisen.

Well, the burden of proof seems to be on you there. In what specific way could the DQA vote tallies ever make a significant difference? Can you come up with a non-contrived example? Even if you can, it seems highly unlikely that it would be very common.

In 99.9% of cases, by far the most relevant data for assessing the location are the current observation record details, the community ID, the comments, and the observer’s observation history (i.e. what they were doing at around the same date/time as the current observation, and/or their general pattern of observing).

The current evidence should be assessed purely on its own merits. The perceived motivations and competence of either the prior voters or the observer should never form any part of that. That would risk subjective bias entering into what should always be an objective, scientific assessment.

The opposite seems more clearly the case. The previous votes are a source of actual confusion, since they don’t relate to the current state of the observation. Resetting the votes is an efficient way to speed up the process of removing the casual status of an observation, which in the large majority cases will no longer apply. It’s more productive to focus on that positive outcome than be dictated to by hypothetical bad-faith actors.

In general, the interface should always try to work in favour of those with good intentions. This is the default assumption enshrined in the DQA for every new observation: the votable items all start out with green ticks (but no votes). Thus, when someone edits the location, it’s more consistent to restore the default state.

As suggested by several people earlier, the best way to preserve the context of prior voting is to leave appropriate comments. By themselves, DQA votes are always somewhat ambiguous, since they don’t carry any explanation of exactly why they were made (which is also a weakness of editing histories).

16

I think this feature request only works if the notification portion of it is also implemented. Clearing the votes silently means that the observer could fix the observation in a way that is still incorrect and nobody would know until people organically come across the observation again.

Overall, though, I think this is very necessary – having to leave a comment to the effect of “hey let me know when you fix this observation” every time you vote against the DQA criteria isn’t foolproof (since it requires the observer to do extra work on their end too) and kinda indicates to me that there’s a flaw in the site itself that needs to be addressed here.

I just submitted another feature request without seeing this one that applies more broadly and without the resetting portion – this same thing should also apply to the “Date is accurate” and “Evidence related to a single subject” criteria, since they’re the other DQA criteria that the observer can easily fix themselves. (I interpreted the “no feature requests about notifications or the dashboard/homepage” to mean how the notifications feature works and not specific things that should trigger notifications – my mistake if the latter also falls under that.)

17

I agree. I’ve largely stopped using the date/location DQAs because:

a) It’s often a suspicion, and I don’t have time to get into a conversation with the observer to verify

b) I doubt that many of the observers will actually make any corrections

c) I’m concerned that iNat will implement something like clearing the DQAs without implementing notifications.

For the most part, if I suspect the date/location are inaccurate, I just mark the observation as “reviewed” and move on. If it’s a regular contributor who appears to have just slipped up, I’ll take the time to ask. If I see a pattern with a particular observer, I add them to the list of observers whose observations I filter out. No time to constantly second guess whether date/location is correct.

18

Yeah, most of the observations I’ve come across that had an inaccurate location were glitches or mistakes where a coordinate got reversed, the observation was put in a random spot in the ocean, or the observation was put at (0,0) off the coast of Africa. In those cases there’s not much uncertainty about whether the observation was accurate, so I still use the DQA voting, but I agree with your point and in many cases I prefer to give the observer the benefit of the doubt without strong evidence to the contrary.

19

I was thinking of cases where the image is obviously a phone snap of an image displayed on computer screen or the back of a camera. If the location pin is precisely on a house, it’s almost certain that the location is not where the original photo was taken.

Also applies to a lot of specimen photos. You’ll see a ragged dermestid eaten specimen with yesterday’s date…

I’ve tried asking if the location/date is correct or not dozens of times, but I typically get no response from the observer. So now if I suspect the location/date are incorrect, I usually just click “reviewed” and move on.

There’s a whole other class of observations where I’m fairly certain the observation was not made at the location indicated. There are groups of observers that I know are typically in the field together, and I recognize when all the observers photograph the same organism, but one of the group has placed their pin away from the others (typically, right on where google maps displays the name of the park for example, in a spot where I know the plants shown in the photo don’t grow).