Influx of observations, trust in suggestions feature, extra curation needed!

I love the new autosuggestion feature of iNat -- I keep reminding myself that it's 'learning' as more and more species are added to the algorithm... In the meantime, I've noticed quite a few observations (just in my area of Dallas/Fort Worth) that have been ID'ed as things that area way out of the natural distribution.

As the city nature challenge gets closer, I predict that there will be an influx of observations from new users that trust solely on the visual suggestions given by the app.

I also know that there are many students that are forced to use iNat for some project under duress, and many of them may not care about the quality of their observations... I wish it were different, but I do think that's the reality of it!

Most of all, I'm extraordinarily grateful for the folks that devote their expertise and knowledge to adding in ID's and curating the observations. Know that each time you add an ID, you're actually welcoming someone to the naturalist community -- validating their observation and letting the observer know that we value the observations in the database (even if they're super blurry!). :)

So, I'm curious -- what's the best way to deal with the influx of new observations? I try to add in ID's, but I tend to focus JUST on my region of Texas -- I'm sure there are lots that are popping up from other parts of the world that need the same or more curation... I also use some copy and paste messages, but I may have to modify them a bit...

What do you think?

Posted on 15 April, 2018 16:56 by sambiology sambiology


Three easy steps! Quit your job. Avoid any other hobbies. Abandon your family. I've managed two of those three, so far. And I've been focusing mostly on certain taxa within a fairly broad region (i.e. 500 km radius of my home for Heteroptera and Anemones, adjacent states for Bombus). Some species within those taxa I'm not yet familiar with (or can't easily be IDd with pics), and I just ignore those. If I learn something new which is easily identifiable, I'll do a search for those in ID mode and confirm IDs.

Posted by pfau_tarleton about 6 years ago

I'm guilty of those visual-based observations in regions I'm new to, though at least I'm self aware enough to step it back to the genus level (usually)! I sometimes wish there were regional curators like on ebird, if only to say, "LOL, definitely NOT that!" But there seem to only be enough of those in urban areas or ecological hotspots. I'm just crossing my fingers that one day a plant nerd will come through and correct all my rural OK observations :)

Maybe if there were a way to gamify IDing others' observations - badges, special honorifics, exciting sound effects...

Posted by marissa-b about 6 years ago

It's not what you meant, but I've been curating some taxon photos to put better overall images, leaves, discerning features, etc. toward the front so they show up in the autosuggest "compare" photos. I was comparing two flower species and both had tons of flower pics but not the leaves which would make it easier to ID. I know new users won't know what to look for, but every little bit helps, I hope. PS. Yesterday I visually ID'ed an "easy" one as the wrong thing and I'm thinking some days autosuggest is much better than me, even if it does think Chuck is a trout and James is a bat.

Posted by kimberlietx about 6 years ago

@marissa-b PS. I think it's the Discover Life guides that uses crowd sourcing to help categorize submitted images. It's a game site... like "Click the images with moths that have spread wings", and then they use the ones with a high confidence.

Posted by kimberlietx about 6 years ago

I'm guessing that the expected influx of observations during the city nature challenge will be handled by the usual band of inat local yokels present in every major city area. At least in Washington and Oregon there is the regular bunch of submitters & ID'ers. Hopefully we can handle the uptick in observations and provide welcoming identifications to new folks joining inaturalist. We'll argue about whether it's an Olympic gull or a pure glaucous-winged gull or western gull. And if I had a dime for every western sword fern I've identified, I'd buy a nice microscope. As far as the new autosuggestion feature, I love it, too. It is amazingly accurate. There are things, mainly plants that I've seen before, but have no idea what species they are. For organisms where there is some uncertainty or ambiguity, I can usually narrow it to one species or to genus by looking at the range maps, researching online, going to my trusted field guides, and taking into account my field observations (is that peep running around a sandy beach or foraging on a mudflat?). I think we just need to spend a little more time ID'ing, especially May 1-3, which is when the ID'ing takes place for the city nature challenge and then the results come out May 4. I'm rooting for Seattle, naturally (all puns intended).

Posted by brownsbay about 6 years ago

I must have a different version of iNaturalist--I haven't seen any "auto-suggest" feature. Maybe I don't know where to look?

Posted by a43560 about 6 years ago

I wonder if the auto suggest which I like could only suggest those in the range? For instance when I post something that is endemic to Texas why would it give me results in Florida only. I also have noticed that there is visually similar and visually similar and nearby which helps me. For the insects I do, it still has a long way to go and the plants are hard, but it is better then nothing that we had before for newer users.

Another option is a probability or just going to Genus or Family which I think @tiwane said it was not set up for yet.

Anyway .. We will have challenged in the inat challenges and then improve from there :)

Posted by butterflies4fun about 6 years ago

@sambiology-as an iNat newbie, I am as thrilled about this community as you. As an old dog with some biology background, I have put in a few observations that I was "most definite" about since the nature walk at Mineola Nature Preserve 4/14/18. The autosuggestion has been correct most of the time. On a couple, it threw in some suggestions out of left field...probably due to my photo. Mainly, I am so excited by other members noting on my observations even if it is an agreement.
Your comment: "Most of all, I'm extraordinarily grateful for the folks that devote their expertise and knowledge to adding in ID's and curating the observations. Know that each time you add an ID, you're actually welcoming someone to the naturalist community -- validating their observation and letting the observer know that we value the observations in the database (even if they're super blurry!)"... I agree with wholeheartedly!

It does help tremendously for those of us new to the group and who are eager to learn!

Posted by charlottewatsonsa... about 6 years ago

I've been surprised to find a good amount of accuracy in the autosuggestion feature. Sometimes, it's just what I need to prod my memory to ID the species. But there are times when it has been way off in left field. As someone in previous comments suggested, it may be because of my photo not containing enough information. In an attempt to add details in order to get an accurate ID, I probably post too many photos. But I would rather see too many than not enough. I wish we could get iNatters to include more than just the flower on a plant, for instance. Not only would it help with the ID, but it would help create data information for the autosuggestion feature (as Kimberlie has suggested). As a curator for some projects, I just pass up observations/don't spend valuable time on observations which don't include enough for an ID. That includes the blurry photos. Sam, you might repost your journal writing about things that we might add to help the iNatter with such an observation. You are so diplomatic!

Posted by suz about 6 years ago

I am fairly new. I started last April 2017. I too have some blurry photos. I am thankful to the auto suggestions. I forget the names of some of the strange tiny things I come across. I am not sure which category to put them under, sometimes. I love to be apart of this wonderful community of Inaturalists. I Have noticed when uploading some things in the past. I have had auto suggest, suggest something not in my area as I found out from some identifiers. I still am finding things I have not seen before. Autosuggest has been helpful, it also has messed me up. I still like it though. I also am thankful to the Identifiers that help me out.

Posted by walkingstick2 about 6 years ago

I've been doing my best to always put the location in before going to the ID field. It seems like it does better when it has an area to work within. It also makes me feel a little more confident in my original guess when it's been verified nearby.

Posted by fratto about 6 years ago

@butterflies4fun The autosuggest thing relies on a large sample size of photos to base itself off of. Inat staff are aware of the issue with it suggesting species out of range, but when you think about how each individual species would need to have distribution data entered and checked... I think the task is very daunting at the moment. For the most part, I think it works great, and the big hiccups are addressed by the individual users going around doing IDs.

Posted by nanofishology about 6 years ago

I think iNat's Getting Started series has a decent set of suggestions on how we all can become ID'ers, so I am trying to learn a few things in this area that I can confidently ID.

And my next comment is a little off topic, but it seems like a problem in growing iNat - I have been encouraging my sisters to post to iNat, which they do, but then their observations are never ID'd. They're both good students of nature, but have the misfortune of living in areas that are not well represented on iNat. It means the Auto-suggests aren't very strong, but also that there's no one looking around for observations to verify. Even in Texas (one sister is in Kimble County), where the TPWD supports iNat, it's a challenge. It's so satisfying to get an Observation verified, but when it never happens, the posters will begin to lose interest. And maybe there's something I'm missing in how to get posts reviewed in these less dense iNat areas.

Posted by apcorboy about 6 years ago

Because I have a large number of students that frequently use the auto-ID feature I always try to make a pass through their observations and at least put them on the right track. That may not mean getting it to species but sometimes it is just getting it into the right kingdom. I work under the notion that every step towards the right identification counts even if it is at a base taxonomic level. I also do this because I know it will throw that observation out to someone who might be interested in that group and will more likely take a closer look at it. While I think that auto-ID feature is great it does have its drawbacks as noted here but it also has the drawback in that it doesn't really encourage someone (especially a new person) to go searching for the answer and maybe learn something in the process. I often ignore the auto-ID suggestion myself and try to chase it to my own satisfaction because I feel that is the best way for me to become even more familiar with the organism.

Posted by cmeckerman about 6 years ago

Ok..I am one of those New people you mentioned! I have slightly blurry pics. I want my pics to be very detailed & crisp but I can't quite figure it out on all my camera settings 😑
Of course I made sure it's on best quality...I tried zooming too, HD on & off, shutter speed & I'm slightly technical but I'm sure there are people joining that are worse off than me. Maybe a basic check list for camera settings for newcomers would be helpful. In the meantime I will keep playing with my Android camera.

Posted by lara38 about 6 years ago

Add a Comment

Sign In or Sign Up to add comments