This is not really a bug report, because this is clearly working as intended: when I add too many observations at once (over a hundred) I run into some kind of “overload protection” and I will not get AI suggestions anymore, unless I pause for a few minutes. This is unlikely to affect people who upload from their cellphones, but I want to take high quality macro images with my camera so that the insects are clearly identifiable - and then upload them all from home. Sadly, the webpage wants me to observe less :(
3 Likes
Just wait until they all (or most of them) are loaded, I upload 200 pics and AI is working (though with macro you better add own ids, AI will suggest wrong taxa in many cases), it just won’t work if all resources are going to uploading photos. Anyway staff says to not risk and not upload many observations at once, and 100 is many.
I never ran into this because I prefer to work in smaller batches (maybe 10-20 observations) anyway. Would that be an option (and easy way around the issue) for you?
6 Likes
The behavior is really consistent with deliberate throttling, because it consistently works to wait for a while and then I get some “allowance” of AI requests before it stops again - it wouldn’t be behaving like this if it was just a problem of uploading the photos, it also doesn’t kick in until I do like 50 photos, even though at that point, they are actually all uploading. I guess doing small batches would introduce some downtime that hides it, if you proceed slowly enough …
I am aware that the AI gives wrong IDs, but there are plenty of insect families where I still have literally no idea - and it gets the family right always all the time, which usually is enough to summon a proper IDer, as many people clearly follow just small subsets of insects, as the order is so huge.
The limitations you are hitting are those of the server. As a free site subsisting on donations and grants, iNat can only afford the costs of a certain amount of ‘server power’, as it were, and restrictions on the amount of load one user can put on the system are in place to ensure the site doesn’t go down or slow for everyone else. I’m sure iNat would love nothing more than for you to upload more observations, but other than giving iNat the substantial funds to increase their server bandwidth, I think all you can do is keep your upload batches to 10s and 20s rather than 100s at once.
11 Likes
It is connected with you uploading many photos, uploading 5 at once doesn’t need same bandwidth as 50.
When the “new” uploader was first introduced I got burned a few times by working on a large batch of observations - selecting locations, working on ids, adding notes, etc. Then I would hit submit and sometimes get an error message and lose them all! So I just got in the habit of working in smaller batches which works well.
4 Likes
Does sound like a server limitation. (iNaturalist itself wants you to post lots of observations!) I’ve posted 50 to 75 at a time without problems, but you post a lot more at once!
Interesting. I have never had this problem although I often upload 100 or 200 observations in one go. However, I usually try to put at least a very rough ID on them before I upload them. Then a bit later I go back and try to refine my IDs, taking my time.
My computer freezes if I do more than 30 or so at a time! Ha.
4 Likes
Smaller batches seem to work best. Even doing only 30 at a time, I often go do something else off-computer for a few minutes while it uploads as nothing else will work on my computer while it is uploading.
… when I upload over 100 at a time, I usually add all the photos, then walk away from the computer and do something else. When I come back 1-3 hours later, I do the locations, labels, IDs and click “upload”. But I agree, it looks like the system tries to keep you from uploading more than 50 at a time…
I regularly upload 100-150 photos at once, sometimes more, and have almost never encountered any issues
1 Like
Since I don’t upload a lot of observations, I cannot comment on that specific issue. However, if I want to go to Identification page 999 (for example) to look at older observations, I cannot do that. As above, likely a server constraint.
It’s kinda related, as I understand it it is manually set to not load more than 10000 records as it seems all of those pages are loaded when you choose it?
1 Like
I load in batches of 45 to 60 at a time. Then I go through and update all the locations , using the map to pin-point and correct slightly off GPS coordinates; then I will go through using the AI (or not depending on how accurate it is), this will be for approx 20 at a time as I only do what I can see on screen and then when those are done I scroll down to the next set in the upload. I never have any issues with overloading the servers. Hope that helps.
1 Like
Whatever the reason, as a mainly IDer, I am glad it does this. It is really frustrating to hit a wall of a dozen to twenty pages of one observer’s observations of a taxon or group that I can’t help with. Frankly, it feels rude. If that same number of observations was interspersed with other observers and/or other taxa, then it wouldn’t feel like someone was hogging it all.
But there’re filters made specifically so you won’t see groups you don’t want to id?
1 Like
I don’t quite follow why someone uploading a lot of observations is ‘rude’
2 Likes
I just explained: dominating twelve or twenty pages so that nobody else can get a picture in edgewise is rude. Please be considerate of IDers who might not be able to help with your particular taxon and would like to do something besides “mark all reviewed” for twelve or twenty pages straight.