We’ve taken some steps to try to limit machines masquerading as humans on iNaturalist. There has long been a norm on iNaturalist that content should come from humans whether this be observations (e.g. a human pointing a camera at an organism or a human selecting from a bunch of images taken from a camera trap) or comments and IDs (humans choosing which species or authoring remarks). But lately we’ve had increasing issues with machines masquerading as humans on iNaturalist causing an outsized impact on the community, and we’re taking steps to rein this in. We made iNaturalist for humans. When machines masquerade as humans and make content on iNat with no human oversight, bad things happen because our software and our community simply were not made to accommodate the speed, volume, and other characteristics of machine-generated content.
Specifically, we updated the community guidelines to include machine generated content as a suspendable offense and tried to describe in detail where we are drawing these lines.
Likewise, we made it so that users must apply and receive manual approval for permissions to create API clients capable of posting authenticated requests (i.e. apps or scripts built on iNat that facilitate the posting of content). This won’t impact existing API clients, just new ones. The application should describe the purpose of the application (e.g. ‘An app I want to make to help me upload my own observations from my Desktop to my iNat account’) and we’re requiring that accounts are two months old and have made at least 10 improving identifications in the last month to be able to submit an application.
Thanks all for your patience as we try to keep things as open as possible while maintaining the integrity of the iNaturalist community!
If you have questions or concerns about this guideline you can bring them up in this topic, but please don’t speculate about accounts that may have used (or are using) machine generated content. If you believe an account is using machine generated content, please email help@inaturalist.org so we can investigate.
Not exactly this, but I’m curious if there has been any significant issue with machine-generated accounts (not just content)? If I’m reading this right, it would only address content generated by an existing account, and not whether an account itself is machine-generated, correct?
That is rather unfortunate and excessively broad, IMO.
I’ve been thinking for a while that it would be quite helpful to have some form of bot(s) that could simply link to ID guides for certain difficult taxa. Given the ever-growing backlog of observations in the ‘Needs ID’ pile, assistive tools like that could be helpful.
Can the underlying issue —
our community simply were not made to accommodate the speed, volume, and other characteristics of machine-generated content.
— be addressed through rate-limits or similar technical fixes?
i’m not sure the stuff in the parentheses is (necessarily) the same kind of stuff as what you would actually be limiting with this change. i’m not against the change, but i’m just surprised that folks who wanted to post a bunch of machine-generated content would have done it by going down the path that you’re limiting here. maybe you all are taking other measures to close down other paths for machines to generate content though…
I just noticed that I can no longer edit my oauth application record that I created a year ago. I get a “You don’t have permission to do that.” error on clicking the Edit button.
I haven’t yet applied to be an “App Owner” - so I won’t have the new app_owner role. Is this change intentional, or should editing old applications be permitted without applying to be an App Owner?
agoranomos if you can submit a brief application we should be able to proceed quickly. I should have been clearer that creating and updating requires this new app_owner role