Timed Suspension Discussion

[upupa-epops]

From what I understand, “machine-generated content” has historically on iNat referred to content that is produced with automation, e.g. running a program that uploads hundreds of observations without a human supervising every action.

[cthawley]

Yes, I agree. The definition on the suspension page is:

“Machine generated content: We do not allow machines to generate and post content on iNat with no human oversight curating each piece of content, and any account suspected of doing so is subject to suspension and the removal of the content. Read more about what constitutes machine generated content here.”

I think this is considered differently from the AI-generated media flag that @jenna489 was referring to (and much more seriously). I would think that AI-generated stuff might have the same suspension progression as Copyright Infringement? But I agree that it should be explicitly included on the chart.

[karthik_83]

I too agree. I didn’t read the text under ‘Machine generated content’ in the new guidelines, like most others here, to also include AI generated media.

My sense is that it is a lot less common, and most users/curators don’t come across full-on machine generated IDs, scripts, etc. AI generated media meanwhile is (and will become) more prevalent.

So, it’d make sense to mention the guidelines for it explicitly (or mention/combine with copyright violations).

[jenna489]

Thank you! Given that the purpose of these new guidelines is (at least in part) to enable more consistent treatment of violations, making explicit the expected treatment of AI-generated images would ensure that we are all on the same page and well-equipped to apply consistent treatment.

[jenna489]

@cthawley

Can we clarify if there is/isn’t crossover between the infraction categories? For instance, if a user gets two minor insults (and thus a one day suspension) and then gets a major insult would this be a one day suspension (first major insult) or a one week suspension (which is what a third minor insult would get)?

Agree that clarification on this would be much appreciated. CNC, at times, felt a little bit like whack-a-mole — e.g., a user uploads copyrighted images → warn them not to do that —> they start uploading AI-generated images —> warn them not to do that —> they start falsifying dates → warn them not to do that…etc, etc.

[cthawley]

Sure happy to try to clarify, thanks for asking!

For

@cthawley

A dropdown for “Infraction” that allows a moderator to choose some of the same things as the suspension dropdowns would really help keep this organized and clear.

I wasn’t necessarily proposing a free text option like

@caseyhh

• Add a field to track curator’s comments on a Moderation Action, which is displayed to other curators, not to users.

though that could be good. All the things I was thinking about would be on the Moderation History alone which would only be visible to curators. Currently in the “Body” column, there is a moderator note or the text of any flag, either the user-entered flag text or the standard flag text. In cases like “copyright infringement” or “spam,” these auto-generated texts are helpful in assessing user history, but there is also “inappropriate” which is not, as it could really be anything. For flags with user free text, it also isn’t necessarily clear which (if any) of the infractions from the Suspension Guidelines may be applicable (sometimes the text of the flagger doesn’t make much sense or is wrong for instance).

My proposal would be to allow curators to add an optional value either in the “Body” column or another, new column for “Infraction” from a dropdown of (almost) the same standardized choices as the “Reason for Suspension” dropdown. However, I would want the insults choices to be changed from “Insults or Threats” to be broken out to the choices in the Infractions table (“Minor Insult”, “Major Insult or Threat”. This would allow Curators to quickly scan the Moderation History page to get a view of any past issues a user has had. This wouldn’t overwrite the original value, but just be underneath it (shouldn’t be an issue as the lines on Moderation History already expand to accommodate longer texts)

For

@caseyhh

• Add a field to suspensions indicating if the user was warned BEFORE they were suspended.

my proposal is a bit different.

I would propose adding a column that is “Warned?” or “Warning” or something like that which notes whether a warning was given by a curator for that specific action/line in the Moderation History (not any suspension). Curators would have the option to select a yes/no value for it to denote that a user was warned about a given action/line in the moderator history. So the warned value could be on the entry for a flag for copyright infringement or whatever. Combined with the Infraction value proposed above, this would give curators an easy way to scan the history and say “Ok, this user has 5 flags for copyright infringement but they were warned only once”, so a new copyright flag would just lead to the second offense (Suspend for three days).

This would also be useful because sometimes currently a moderator has to add a moderator note noting that they warned someone for a previous action on the moderator history. This adds another line, and it also isn’t always clear exactly which issue/flag/line of the moderation history the warning is for (usually the one preceding, but not always).

For the Action column, I don’t know all the permutations possible, but it only seems to have “view” in it and the view text links to either the flag itself or the flagged content (like a comment or identification) when it is hidden or unhidden. This is already a little inconsistent, and, when it links to the flagged content, just duplicates whatever link is already in the Type column (so it is unnecessary in that regard).

My proposal would be to just make the “Flag” text in the “Type” column a link to the Flag (like it already is a link for the content - user, photo, message, comment, ID, whatever). The “Type” column could then go away and free up space for a column in the moderator history page to use for something else like the proposed columns above.

[tiwane]

Just a heads up that @caseyhh is an iNat staff member, I just added him to the staff group on the forum, so “iNaturalist Staff” now appears next to his username. Sorry for the delay on that.

[caseyhh]

Thanks @tiwane !

[tiwane]

Thanks for all the feedback, it’s much appreciated! I’ll try to respond to some of it here. FWIW, we haven’t received any appeals from suspendees of timed suspensions, as of earlier today.

---

@jenna489

I have one question. I don’t see AI-generated images among the list of infractions — would repeatedly submitting such images fall under “Continued activity that reduces data quality (such as adding many poor IDs)”?

Yeah, that’s what I was thinking, at least at the outset. But we could maybe combine it with the copyright infringement category as well.

@jdmore

1. Most, but not all, of the Infraction types specify at what stage any Curator status of the offending user should be revoked. I would like to see this listed consistently for every kind of infraction (even though it’s something only Staff can implement). This is currently missing for Harassment, Copyright infringement, Intentionally adding false content, and Continued activity that reduces data quality. (imho, any of the infractions listed in the table indicate to me that someone is not yet ready to be a Curator on iNaturalist.)

Thanks, it’s an oversight and I’ll add it to more places.

@schoenitz

• First, second, etc. incident for copyright infringement might need a workable definition. Users sometimes upload 10-20 or however many copyrighted images at a time. They may not know any better. That probably shouldn’t count as 10-20 incidents. So how do we count then? Days? Maybe after 1st, 2nd, etc. warning? That would need organized moderator notes, as jdmore notes.

I was thinking after the warning, and giving them a grace period to make sure they’ve had time to read it. So, Infraction discovered, could be after 1 stolen photo, or posting 50 → warning given = first incident. But agree it needs a more clear definition.

@schoenitz

• There are some curators now active who make an effort to talk to users as they would to equals, ask questions, offer help, and the threat of suspension is no longer the first thing that comes up. In my experience here and elsewhere on the internet, users generally respond in a more productive way to this.
  Looking at the How to suspend an account page, I’m concerned that the actual moderating is going to decrease again and mechanical application of all those rules and tools will increase. It’s going to feel unwelcoming for a lot of the less experienced users.

Yeah, it’s a tough thing to make part of a universal process. Some people might want to engage with a user before a suspension, but others may not, or may not have the skills or time to.

We decided to go more toward a shorter first suspension or warning system for minor/non-malicious infractions, eg minor insults, copyright infringement, so that the consequence isn’t too terrible (like being suspended for a day, or getting a warning) so that it’s not a huge deal or something a curator might waffle over, that will still let the user know what’s not acceptable without them being affected by a long suspension. But we’ll have to see how it plays out.

@eyekosaeder

One small question: What is the difference between an indefinite suspension vs a permanent suspension, and an account ban vs a ban from iNaturalist?

See https://help.inaturalist.org/en/support/solutions/articles/151000232621-how-to-suspend-an-account#Intervention-definitions I think they’re fairly clear, but if you have specific questions please let me know.

@insectobserver123

I think wherever possible we should prioritize deescalation over punishment, and would prefer to see the suspensions for minor insults be treated more as maximums than defaults.

I just want to make sure that these aren’t thought of as “punishments” (I don’t meant to pick on you, just using this as an example). Let’s really think of warnings and suspensions as interventions to stop/prevent behavior that is detrimental to iNat. We are not trying to inflict pain (which “punish” implies), just trying to protect the broader community and uphold the guidelines/ToU.

@insectobserver123

My biggest concern is the conflation of major insults and threats. A threat is defined here as “any content that indicates an intent to harm another person”. This is a fundamentally different thing from an insult, I have a really hard time seeing how a 1 day suspension is appropriate in any case where someone has stated intent to IRL harm another user

Noted, I’ll take a look at separating.

@insectobserver123

Well said, revoking curator status based on the accumulation of 2 incidents of behavior, even if that behavior is neither serious nor the person’s typical behavior, could end up functioning as a limit on the duration of curatorships, leading to less experienced curators (and less curators in general).

FWIW, for years now each curator is told “We ask that our curators be models for other iNat users, so violations of our Community Guidelines can result in loss of your curatorial status.” I agree a few minor insults probably aren’t a big deal, but revocation of curator status isn’t necessarily permanent. I should add this to the list of intervention definitions.

@insectobserver123

I was wondering the same thing, I don’t think it makes sense to give a lesser penalty because the insult is worse, so I think the week suspension is appropriate.

In general I would go for the longest suspension in these cases.

@schoenitz

how should we treat cases where a user deletes their observations in order to

The guidelines we posted are for common infractions:

Right now there isn’t a hard and fast rule about this behavior. We’re hiring someone who head up moderation/curation in the next few months, and working on systems that will help curators make decisions and hopefully reduce bottlenecks. I don’t want to set a whole lot of new rules before they come on board, so I think for something like this it will have to be case-by-case for now. In my experience it’s pretty rare.

[egordon88]

@tiwane

We’re hiring someone who head up moderation/curation in the next few months

Sounds great for you and stressful for them :-)

[insectobserver123]

@tiwane

I just want to make sure that these aren’t thought of as “punishments” (I don’t meant to pick on you, just using this as an example). Let’s really think of warnings and suspensions as interventions to stop/prevent behavior that is detrimental to iNat. We are not trying to inflict pain (which “punish” implies), just trying to protect the broader community and uphold the guidelines/ToU.

I agree with this, the point I was trying to make is that in some cases of minor insult, verbal deescalation is the most effective intervention to stop the behavior, and requiring that suspension or the threat of suspension be used has the effect of shifting the guidelines in a more punitive direction than just trying to stop the behavior.

Noted, I’ll take a look at separating.

Personally I think the guidelines for threats should be more strict than those for hate speech. I have no objection to the current guidelines for major insults.

revocation of curator status isn’t necessarily permanent

Ah, this makes me feel a lot better about the guidelines. I do agree that curators should not be allowed to get away with lots of violations.

See https://help.inaturalist.org/en/support/solutions/articles/151000232621-how-to-suspend-an-account#Intervention-definitions I think they’re fairly clear, but if you have specific questions please let me know.

I had the same question actually. As far as I can tell an indefinite suspension is a curator or staff decision to suspend an account until they successfully appeal, and a ban is a staff decision to never grant an appeal, but it is somewhat unclear if this is what is meant

Also the difference between “Account banned” and “Ban from iNaturalist” is not stated, are these 2 ways of saying the same thing? Or are they different?

[cthawley]

In regards to:

@tiwane

@insectobserver123

My biggest concern is the conflation of major insults and threats. A threat is defined here as “any content that indicates an intent to harm another person”. This is a fundamentally different thing from an insult, I have a really hard time seeing how a 1 day suspension is appropriate in any case where someone has stated intent to IRL harm another user

Noted, I’ll take a look at separating.

and

@insectobserver123

Personally I think the guidelines for threats should be more strict than those for hate speech. I have no objection to the current guidelines for major insults.

Direct personal threats of real world harm have been thankfully very infrequent in my tenure. The most common threats I see are for online retaliation, usually on iNat itself, and the single most frequent is some variation of “How would you like it if someone bumped back IDs/cast DQA votes on your observations???” often accompanied by a couple intentionally incorrect IDs or DQA votes. I think the guidelines are appropriate for these kind of things. For serious threats of in-person harm, especially if there are >1 threats in a single incident, I think immediate suspension is warranted to stop the threats. I don’t think all threats are the same, but I think the guidelines as written give the flexibility for curators to make that call.

The scenario above does bring up another question I’ve been pondering though:

What to do in situations/incidents where a user has multiple distinct violations? I’m not talking like 20 copyrighted photos in one day (this type is addressed in the new guidance, and I agree this is really just one incident), but situations where a user does something like make a threat and make intentionally incorrect IDs as in the example above or something else. Conceivably, any suspensions could “stack”, though trying to calculate this could get complex. My personal instinct would be to apply whatever the strongest sanction is (e.g., if it is the user’s second threat, and second instance of discrimination, I’d go with a one week suspension). I would also, however, warn them about both issues, and consider the next time (if any) they did either to merit the third sanction for either.

[insectobserver123]

I was interpreting a threat as a statement of intent to physically harm the person, I’ve only ever seen this a few times, but I have seen it.

Under the current guidelines it appears that these cases would require consultation with another curator in order to suspend for more than a day, which does not make sense. I think threats to mis-ID or otherwise be a pain on the internet can still be handled as major insults, but threats to commit physical crimes against someone, or saying that someone deserves to be the victim of violence, should be an immediate indefinite suspension IMO.

I agree that the current guidelines are fine for retaliatory mis-IDs and stuff like that.