[drdeca]

(I have no information about the specific topic of alleged CTH bots/“bots”. I’m not asserting that any particular group is using bots. I’m just speaking generally about the topic of what counts as “bots” and what doesn’t.)

I kind of feel that in a case where there are many accounts using identical text (which is too long to be identical by coincidence) (e.g. in replies to something), it is probably fair to call these bots, even though one can’t rule out the possibility that the automation is as small as just copying and pasting the same text into different accounts + something to open windows as different accounts viewing the same tweet.

Like, maybe that setup wouldn’t strictly count as bots, but really only because it hasn’t been automated as much as it could be while having the same result.

So, I think it is probably fair to call that observable-phenomenon “bots” even though it might not always strictly be bots?

But I agree that a group of paid astroturfers(/trolls?) who are writing different things, probably shouldn’t be called “bots”, even if they each use multiple accounts (provided that they aren’t semi-automatically saying the same things from the different accounts).

I do think that the concept of sock puppets probably captures much of the idea that “bots” is inappropriately extended to, and in a sense applies to many of the kinds of bots that people are talking about in this context (bots pretending to be many people, rather than a bot which is up-front about being a bot).

So, maybe talking about “socks” would be a good substitute for when some are talking about “bots” when they don’t really care if they are actually bots?

[cwkoss]

Interesting. I do consider accounts managed by paid astroturfers who manually write replies from multiple accounts with different identities "bots" but maybe that is a misnomer. "Sock-puppet-masters"?

Copy-paste reply bots are clearly "bots" IMO - but these seem to be falling out of favor because pointing out many identical replies tends to be a very effective "dunk" on the idea they are promoting.

Someone found screenshots Sally Albright's tooling in a support request: she had ~50-100 different accounts with unique identities. She could post replies to a given tweet from multiple different accounts from a single page. However, liking and retweeting each others replies could be fully automated (with some randomization of timing and which accounts are used).

Albright has claimed that the accounts she managed were originally owned by people who gave her control of them. I am incredulous, particularly because some accounts were identified as using a dead person's image with a different name. But there is the possibility that she was given the impression by whoever provided these accounts to her that these were 'real' accounts where the owners consented to have them used in this manner.

It gives the capability for a single person to seem like they are a crowd of like minded people all in agreement on social media. I believe this model of automated likes/retweets but manual replies is the most common (and pernicious) method used today - practically impossible to prove without data outside the platform it is occurring.

[drdeca]

I suppose if “bot” came to refer separately to mass-sockpuppeting as well as to automated messages, that wouldn’t really be that unfortunate of a change in the meaning of the phrase, so I could accept it.

I would prefer things that don’t have any significant automation to them be called something related to sockpuppets than bots, but sockpuppet is unwieldy because it is much longer, both in speech and in writing, so if there is no abbreviation for “sockpuppet”/“sockpuppet master” which is convenient, easily understandable, and feels right to use, then I guess that could be justification enough to call them “bots” despite there being little automation to it.

Regardless, I think the big issue is how people can use these techniques to massively overrepresent how common certain views are compared to other views, regardless of whether it uses any automation. (Because of this, when I make multiple accounts on some service which I keep separate from one-another, which isn’t often, I try to say in my user profiles roughly how many other accounts I have on the service, to avoid contributing to this problem. This has only come up on one website for me though.)

(Also many-account auto-liking would qualify as “being bots” in the narrow sense IMO even if the tweets were all manual. And automation certainly makes the problem bigger.)

[Jordrok]

Personally, I tend to just call those "fake" accounts. Fake, in the sense that the person behind that account is someone other than who they are claiming to be - whether that is a specific human being with a name and a picture, or just an anonymous, discrete individual who is distinct from anyone else already involved in the conversation.