Add "taxon can't be refined" DQA vote to select unrefined obs. after years


(Link)

Issue: Making most* obs. with unrefined genus, subgenus, tribe, subfamily, or family community IDs become RG after 1-2 years would do much to help review/organize the increasing number of iNat obs. The obs. example shown is from 2015 and has 7 genus IDs. Making obs. like it RG would indicate to others that they’ve (most likely) received the most specific IDs possible, which could save review time for others. This feature would affect obs. by adding 1 DQA-“taxon cannot be refined” vote to each, whether or not they had DQA votes prior (which could potentially cancel each other out). As this would take effect using that same DQA vote system (except in an automatic way), obs. IDs are still free to add, delete, or revise later, which can also potentially change RG status. In suggesting this, I also assume the current DQA option has a useful function in helping to organize/process obs.

How it would work: Most* obs. which remain non-RG (Needs ID) after the time period would get one DQA “taxon cannot be refined” vote added to them, which would have the same effect as an ordinary DQA vote. Eligibility requirements for obs. to be affected would be to have at least 1 ID not from the observer, or at least 3 or 4 total (although that number is open to discussion). Also, there’s a question of whether non-RG obs. with species community IDs should be excluded. I think the core of this request would be fine regardless of how people decide some of the fine details.

FAQs:

  • If implemented, it will still be possible to add or revise IDs to obs., so RG status can change as usual. Many identifiers also already ID both non-RG and RG obs.

  • This doesn’t replace DQA. It’s different since it would affect most* all obs. after a time limit (including obs. where the potential for ID refinement remains uncertain, which don’t ordinarily get manual DQA votes).

  • When does the clock start? Probably when the obs. (and/or IDs) are first submitted, but this could be discussed.

  • Could this also or partly be based on number of reviews per obs.? Maybe, it can be discussed.

This is proposed because adding DQA votes is too time-consuming? I feel like it would create more problems than it would solve. If five people are taking the time to add IDs surely one of them could devote time to voting instead.

If someone adds an ID after a year does it revert to Needs ID? If someone withdraws an ID does that reset the clock? What about deleted IDs?

9 Likes

One year is too short and it will prevent refining ids when expert comes to website and checks only Needs id. If 5 experts added genus id they could vote on it long ago, maybe they’re not even sure it can’t be ided further? One click takes too long?

6 Likes

I do think that some people are reluctant to use the “No, it’s as good as can be” box. Part of this may be humility, but I’ve seen comments on the forum and on iNat that discourage people from using it or are even upset that an observation was RGed at genus level because of it. I personally used to use it more to “clean up” IDs but now use it less because of this. So unless there’s a clarification that we want people to use that box more liberally, then I doubt it is a good solution.

2 Likes

I think I’m with @marina_gorbunova here. I’ve spent the year badgering away at the Eristalis NeedsID pile and even the really old ones that were still at genus haven’t been massively less identifiable - there have been loads of easy ones - it’s just such a common genus, and not super-easy, that observations quickly get buried under a mountain of new ones and not seen again for yonks. Many of the old ones simply hadn’t been reviewed by anyone but the observer.

I can perhaps see the point in making an assumption after say five years and many genus IDs. But I would be cautious about it. There are many taxa that just don’t have many people paying attention to them.

Further to that I don’t necessarily think a large NeedsID pile is a bad thing in itself - the pile keeps iNat ticking in a way. The problem is that interesting species often get buried under a pile of common stuff that just isn’t being done. A blanket research grade would bake that problem in, I think. A cool species in a common genus that few people have gone through will get lost quickly (as currently happens) but then eventually fall into RG at genus without being seen.

10 Likes

I favor the idea of making genus level observations research grade and searchable after a set number of reviews. That way they aren’t defaulting to present for general reviewers but experts can do so to see if they can get any to species

3 Likes

This is different, it would have an effect like checking DQA-“taxon is as good as it can be” for all obs. without needing to mark DQA (many more than can/would be marked manually by DQA even if people were to use it more).

Initial suggestions for these fine details (and this part is also more flexibly open to feedback than the main part of the request): after the time period 1 vote could be added to DQA-taxon is good as can be. That simplicity would help, since we already know how DQA votes affect community ID in various situations. The clock could be based on time of IDs, although it could also just be based on “obs. isn’t RG after X amount of time.” I’ll also update the topic now to clarify some of this.

This would be a larger concern if RG “closed” obs. from future IDs, but fortunately they don’t. Many experts (should) know to also review RG as well as non-RG obs. (if they care about ID quality). So, I don’t view RG as final or permanent. This request is just a way to help categorize/process obs. based on what their eventual community ID is most likely to be. I agree that sometimes experts do but other times don’t know if a genus ID can be further refined further. Since the iNat community ID system would still be otherwise identical, we can also add the species ID (even remove RG) later. Plus, we can also vote on DQA to affect RG.

I think DQA remains a good idea to clean up IDs, that’s part of my point with this request (it’s not against DQA). Part of the “reluctance” of people is because DQA is “hidden,” or some don’t know what it does or initially view it as being useful. Ordinarily for DQA, it should be fine for more people to use if they use it carefully. e.g. “no, taxon is as good as it can be” should be inferred to be true of all identifiers (not only you) if you’re going to mark it. i.e., DQA isn’t applicable to be use for every obs. Where my request differs is that it uses something like DQA for all obs. at the end of a long period (1 year, 2 years?).

I may agree although it seems important that IDs be considered too, and to give enough of a time period to allow time for users like experts to review most of the obs.

2 Likes

I think if the main reason behind this feature request is just the under use of the DQA (not so much the time to use it), we probably could focus more in how observers and identifiers should be using it and how to train users in it.

4 Likes

There’re tons of things users should know, and one new to iNat expert meeting with a group with thousands of observations will likely not review RG in years or ever, for observer there’s a big difference if that happens.

1 Like

It’s possible some won’t, but some will, and 1-2 years can be a long time. Would you suggest 3 years, or that it wouldn’t work at all? If this did go into effect, I assume identifiers would become more aware of it and there would also be more info/discussion to inform users. I ID already-RG obs. part of the time, and find some misIDs when I do. So, the current RG system also has some of the same cons. But iNat has to draw the line somewhere to implement some RG policy, like the current RG and DQA system.

Reviewing already-RG obs. is currently already useful and arguably needed (if time), to confirm or correct each. Also, once someone’s set out to review all obs., RG itself matters less for them, and they prevent themselves from the missing any obs. vs. trusting that all RG obs. are correct.

There is too much of a backlog for this to be a good idea. Just today I went over some of genus Oenothera in east Texas including some from as far back as 2013 (I did about 75 IDs). There are many easily identifiable species in this genus. Not enough people have looked at these observations yet. Also there was a taxon change that sent a lot of IDs back to genus level.

Edit to add:
east texas only
31,506 observations
9,563 needs ID
world
142,857 observations
78,326 needs ID

4 Likes

I’ve updated the request to reclarify that obs. that hadn’t received any IDs except from the original observers would be excluded from being affected, so it would prevent the problem you pointed out. Some standard of eligibility for which obs. would be affected would be used in addition to the time limit, such as having multiple IDs.

DQA and the request partially overlap but are separate. The DQA option would remain as-is (I support the use of DQA when applicable, but only wish it were more easily viewable and quicker to use). The request would just be an additional process which differs in being automatic and applying to many more obs. than user could/would manually mark DQA for.

So if one person added and ID (not the observer) and the observer never left an ID, then it would become research grade? I ran across quite a lot like that so that wouldn’t fix the problem.

Just did a bunch more IDs. I’m seriously sick of looking at Pinkladies (O. speciosa) right now.

Edit: if you have a qualifier of multiple IDs, what about things with disagreeing species level IDs? Are they disagreeing because one group picks the common, and another the correct one? Or maybe neither is correct. Maybe it is an undescribed species. Should an undescribed species be left as needs ID or sent to research grade at genus?

Some of this could still be discussed to determine. To keep things simple, most things can mirror how community ID and DQA votes already work. After the time limit all eligible obs. would have 1 DQA-can’t be refined vote automatically added to them. Possibly the request could be modified based on your comment, to also exclude non-RG obs. that have current species community IDs. Re: undescribed species, I don’t see harm in adding DQA votes to them, although users who note them as being undescribed species also can manually vote DQA “yes,” preventing them from being altered.

Sounds like a terrible idea to me unless it is very strictly filtered. If five people agree to a genus with no disagreements and it hasn’t had any new IDs in a few years, seems like going to research grade would make sense. In most cases, I’d say no though, unless you add a category of research grade observations that need ID. I’ll put this into perspective with what I’ve been doing for IDs recently.

As I’m really busy working on a PhD, I’ve been doing some easy iNaturalist IDs during dinner. I filter for old observations that need ID. I mostly haven’t been looking at anything newer than 3 years old. In many cases, I haven’t been IDing to species if it needs extra research as that would take more time than I have at the moment to accomplish much. I can, however, add tons of IDs to family, genus, or species with very little effort. I find these old observations that need little effort as they “need ID” and are not research grade. I’m sure there is a lot of work that needs to be done on research grade observations as well, but there is a gigantic pile of old observations that are not research grade that are easy to identify. I would say almost none of the thousands I’ve assessed in recent months should be automatically converted to research grade. It might be a different story with insects but I’m working on plants. For vascular plants, most genera I would never put at research grade unless they were ID’d to species. Insects it maybe makes sense to automatically send some genera or families to research grade if they can’t be ID’d from photos but an expert should probably make the decision of which ones should go to research grade.

8 Likes

I completely disagree with this idea in every way.

If you’re not observing in one of the major hub areas (eg. USA, Europe, etc) then it’s common for years to pass (like 6+ years) before even a basic ID is made, let alone a refining one.

Often in these areas the initial IDs are either crude to the point of uselessness (eg. spider) or may well be entirely wrong, but both eventually get fixed as users provide IDs over time. Often this takes a long time and involves a lot of back-and-forth on what the correct ID is.

This suggestion would result in a flood of RG observations that are inaccurate, and would also hinder further identifications to species, resulting a major losses of data and information.

I am utterly and completely opposed to this feature request.

7 Likes

The request’s specified to only be applicable to certain obs. (non-RG, 1 or 2-3 years old, and have 4 IDs). Adding further eligibility requirements is also possible (maybe incorporate number reviews, or exclude Needs ID obs. if their current community ID is species). There are also pros and cons to the current RG and DQA system, and so (any) new modification will also have pros and cons (so cons don’t disqualify ideas). RG is always currently provisional and possible (and if a misID, given to) change. Also, manually voting on DQA currently can change the RG-status of obs. for community IDs ranging from family to subspecies, and this request would merely add 1 DQA vote, after which. obs. would behave as normally.

The request’s like an extension of what DQA already does, but differs in being automatic and implemented more broadly. There are many pros, but also a few possible cons. Those who review many already-RG obs. know there are many misIDs there to correct, so even currently many same issues exist. I doubt this request would result in many problems (beyond the usual), but if it were to we could tighten eligibility restrictions and lengthen the time period. If implemented, experienced identifiers would probably also ID somewhat differently, such as IDing more already-RG obs. than now, minimizing some concerns. Another point is, imagine there never was DQA and people were suggesting it be added for the first time. Many people would have the same critiques of DQA, despite that many currently think DQA is useful.

I think this is a bad idea. I’ve seen lots of old needs ID observations that many people had reviewed that were clearly still identifiable.

4 Likes

Sorry, but I do not see any upsides to this suggestion, and if were implemented I’d likely make a feature request to remove it.

6 Likes

So, if RG is useless, what’s the point?