If I say what I really would like to say, this comment would probably get nixed. I can’t even think of an emoji that wouldn’t achieve the same result.
Yikes.
2 Likes
Autocomplete seems to work by the largest observed or most relevant taxon first, its trying to predict what you might want based on the informtaion it has so far.
Take for example the species Luridiblatta trivittata, long to type out, but is also the first recommendation if you type L trivit(or any additional letters) or luri triv, some species take even less letters
If you dont want multiple options when typing in species as the website tries to predict what you want, memorize some taxon id’s such as 6930 for mallard ducks.
1 Like
How did you read that from what I wrote? I do not upload my observations in the Identify tab.
Uploading one’s own observations is completely different from identifying someone else’s. It is expected that observer’s often use CV and quite uncritically so (and that may include myself for certain taxon groups). That will be later corrected by identifiers.
1 Like
Since I agree to everyone above, I come to the conclusion that we still don’t understand you. Because, from a technical point of view, everything has been explained correctly.
2 Likes
It’s an inherent property of large sets of items, that, depending on the filtering conditions, some will be more easily reachable than others.
Global taxonomic hierarchies contain millions of terms. When searching such large datasets, it may seem liberating to output the complete filtered subset, but it’s generally considered poor UI design to present results in this way. If the user types in the first three characters, and this produces a list containing five thousand items, what are they supposed to do with it?
In the interests of user-friendliness, iNaturalist must reduce its autocompletion results in some way or other. Knowing the exact details of the heuristics won’t make it any easier to use. However, given the imperfect nature of such solutions, it’s often possible to improve on them.
Do you have any concrete suggestions? Or can you at least propose some alternative criteria for reducing the autocompletion results that would make them more comprehensible to the general user?
3 Likes
I think so, because the reactions appear to read something I never wrote.
What I am getting is that, taking myself as an example:
When I am identifying something from genus to species I may have only 3 possible species for some countries.The CV would always give me the 3 species I want in the suggestions.
From you I read that I am not supposed to click the correct species that I know very well from the CV suggestion (because that leaves the CV icon).
Or, you meant to say that people who have no idea what it is just tick the first CV option cause its “similar”
thats both user related. Most others are kinda drifting off to technicall stuff right now…
Not sure if I am correct with this but I can assure you that “using CV” (aka seeing the CV icon) is not always a sign of an inexperienced identifier.
It comes down to being able to argue for your point at the end!
I ID a lot from the app and always using CV because it’s much easier to use. I don’t think that lowers my ability to explain my ID.
I also agree that the most misidentifications by CV are created by the observers, while IDers are desperately trying to fix that.
6 Likes
I think this is because of your first post, which is very easily misinterpreted since you don’t provide any context:
This naturally invites people to suggest legimate counter-examples, and to also question the use of the word “should”.
But it could also be the case that someone used the CV uncritically, and this was then blindly agreed to - but it did not need any correction.
It may be unpalatable to many purists, but this is fundamentally what the CV is for. It greatly increases the chances of arriving at correct IDs with no prior knowledge. If iNaturalist is going to allow literally anyone to suggest IDs, we should be grateful that novice identifiers have sophisticated tools like the CV to help them! In fact, we could go far as to say that such freedoms would be completely unworkable without such tools.
3 Likes
Actually, we are reacting to what you wrote. But perhaps we are not reacting to what you meant. I’d like to clarify what you mean.
You wrote,
Example: I am an IDer. I am good at recognizing a diversity of plants, birds, etc. I often use CV to apply names of species I know because using the CV is easier then writing in the name. You seem to be writing that I should not use the CV. Do you mean that I should not use the CV?
9 Likes
Rightly or wrongly, I assume that what is meant is that identifiers should not use the CV to enable them to identify species they are not personally familiar with.
If that’s the case, I partly agree - too many observers will auto-agree and you end up with an observation at RG based on no actual ‘identifier knowledge’. But I have used the CV, particularly with older observations, to help work out the right direction to move high-level IDs, and I don’t see a problem with that if used with discretion. I don’t go to species unless I have actual knowledge, which avoids the ‘uninformed RG’ issue.
4 Likes
I agree. Our using the CV for identification doesn’t add much to the value of the observation. Of course, many of us don’t use it that way.
I rarely use the CV anymore when I identify or upload observations. This is simply because of personal preference though, and I’ve never thought someone else should have to do the same.
I think I can partly understand where you are coming from if you are arguing that identifiers are prone to blindly follow the CV. It is a valid concern.
1 Like
Yep. Suggestions (in vain, as it’s against the very raison d’être of iNat): (a) transparently disclose the rules that are ordering stuff at any point (e.g. “our suggestions ranked by visual proximity above 30%” “suggestions ordered by popularity with people in your city” “three suggestions at random according to your text input”); (b) try and reduce unwanted machine-induced nudges in human decision-making (e.g. back off under certain conditions, like below a known threshold or at the flip of a switch in the user’s settings).
1 Like