Thread by @macg1llivray

Folks, it's time to talk about how to deal with bots on Twitter, strap in. 1/n

#nspoli #FluTruxKlan #FluTruxKlanGoHome #cdnpoli #ableg #onpoli #KarenKonvoy #tonkatantrum

The foreign-funded insurrection we are witnessing is backed by a lot of the same methods we see in the states - bot accounts being used to get hashtags trending and create the appearance of broad support. 2/n

Many of these accounts were either started in the last six months, or are old, abandoned accounts which were compromised in a password leak. In either case they tend to have few followers, and many have very few tweets that aren't replies. 3/n

Twitter, like many SM platforms, algorithmically rewards content which generates interaction, and punishes those which don't. This is the initial barrier to using bot accounts: they haven't been engaged with enough to be widely seen. 4/n

Bots need to generate interaction to be seen; they do this by trolling, of course. Usually in the form of replies to popular threads which they wish to discredit. This is part of Steve Bannon's "flood the zone with shit" strategy. 5/n

vox.com/policy-and-pol…

Bots/trolls will bait you into interacting - specifically replying - by posting weak, easily refuted arguments on purpose. As a decent person, you don't want to see falsehoods prevail, so you reply. 6/n

Tweets which get no replies or likes will frequently be kept "below the fold" in the "Show More Replies" accordion below the more-engaged tweets. Interacting with bot tweets brings them up above the fold where they're seen by other users. 7/n

Once the bot has baited you into interacting, it gets moved up above the fold and is seen by more users, who in turn interact even more, compounding the issue and giving the threat actor behind the bot exactly what they want. 8/n

How do we stop this? There are a few techniques.

The first thing to do is block early, block often. Merely muting a bot doesn't prevent them from using your thread to boost their message, it just prevents you from seeing that it's happening. 9/n

This can be done pre-emptively, too; if you see accounts in other threads with the telltale signs of a bot - low followers, generic account name, canada flag or other generic profile pic - block them right away so they can't use your tweets to generate interaction 10/n

Another technique is to govern your own interactions with bad-faith bot tweets. Does the offending tweet have no likes or retweets? Don't interact! Don't reply, block the user and don't acknowledge them. 11/n

If you feel absolutely compelled to rebuke a bot - as I often feel - screencap their dumb tweet and post it on a new thread. It doesn't count as an interaction and you can use their deliberately wrong post to discredit the entire movement. 12/n

Twitter also allows users to limit replies. I have been scornful of this feature - I'm not here for a one-way interaction - but we are trying to stop an insurrection here, and limiting replies means only likes and retweets can happen, nerfing the bots at the source. 13/n

You may also need to guide your followers away from interacting. DM them if their DMs are open and suggest they delete their interactions. If their DMs aren't open, mention them in a tweet outside the thread and politely encourage them not to engage. 14/n

Having become even peripherally involved with the resistance to the #FluTruxKlan, I'm seeing a huge increase in the amount of inauthentic replies to my tweets. I've been active in trying to suppress them, and you can help too by following these guidelines. 15/15

#nspoli #cdnpoli

Preemptively block them if you see them follow you, you don't owe them an audience. They're following people because many users will immediately follow back anyone who follows them, especially new users looking to build their network.

https://twitter.com/douglas_darby/status/1491102219714056193?s=20&t=aivgBc4y2DN7EcTUC1jCRg

lol, this one even says "burner" right in the username.

This kind of opposition demonstrates that these techniques threaten their efforts. Let's get to work.

#nspoli #cdnpoli #FluTruxKlan #OttawaOccupation

Tools like @BotSentinel can be helpful in determining the authenticity of an account, but the rule still applies: if the tweet sucks and has no interactions, leave it that way.

https://twitter.com/LesLore1/status/1491104935861448704?s=20&t=aivgBc4y2DN7EcTUC1jCRg

Update: they mad

We know that means it works.

#nspoli #cdnpoli #FluTruxKlan #TonkaTantrum #OttawaOccupation

Having scrolled briefly through the aforementioned oberweinermeister's tweets, it's likely the "9" in the username indicates that this is the 8th time this user has recreated their account to evade a ban.

Tell me you didn't read the thread without telling me you didn't read the thread! Literally actively increasing the reach of a tweet which you are also filling your diaper about.

Today we start changing bots' pronouns to "was/were".

#nspoli #cdnpoli #FluTruxKlanGoHome

Here's a good example of baiting: a mocking tweet, with a trite, boring reaction gif and an exhortation about "logic". This actor is trying to create the impression that they want a good-faith debate by using that word, but they're just here to discredit what I'm saying. Bye!

no no I am very real canadian from canada and I am wanting my freedom rightses and these are manys of us

(j/k this is a french-canadian sockpuppet)

I have been conflicted about anonymity in the past, but in reality it's difficult to enforce at scale.

If everyone's going to have the option of anonymity anyway, I'm glad people can use it to have a safer experience on the platform.

https://twitter.com/playlearnlfm/status/1491167033035087876?s=20&t=jhgEok_EjO6OmHUWRqdSjw

A name like @playlearnlfm is not in and of itself indicative of someone being a bot or troll. Lots of users have usenet-style usernames - like a short nickname with no spaces - or an obfuscated/created name for privacy.

lol, lmao

the above user was suspended before I could even screencap it!

I really want to do another thread on profiling bad accounts on twitter. There are definitely different types!

https://twitter.com/Lee98604032/status/1491177405469577216?s=20&t=LfJz5-xEwAs4Bt_IXdH6jA

Here's an example of an anti-vaxxer quote tweeting me to perhaps get an engagement. The tweet *looks* like good-faith if you don't look at their profile, but it's a bunch of anti-vax nonsense if you do.

Tell me you didn't read the thread without telling me you didn't read the thread part two!

Quote-tweeting content you absolutely loathe to own the libs. These people really think they're going to win, eh?

True automation is actually somewhat rare, many trolls at scale are humans in troll farms using accounts which were co-opted by an automated or manual compromise process. This is somewhat remarkable, that there's coordination of automated processes.

https://twitter.com/Mokinokaro/status/1491188761132412928?s=20&t=GuAhvyYnoA6C4ftX1SiFrg

tfw honk honk no change guvvermint 😂😂😂😂

#FluTruxKlan #nspoli #cdnpoli #OttawaOccupation

I read this one wrong! Antivaxxers frequently use the terms "immune system" and "tyranny", and I saw those terms together in short order in the context of a paragraph about cytokine storms which I was clearly out of my depth trying to read. I apologize!

https://twitter.com/macg1llivray/status/1491180496583557124?s=20&t=jhgEok_EjO6OmHUWRqdSjw

So what else did I get wrong here? Maureen's real name is her username and the photo does not look generic, she did not have any other red flags other than occurrences of the two terms.

Her stats and tweets suggest a long-time, casual user.

So really, I needed to follow my own rule: no one criteria is 100% diagnostic!

Here's a great thread about some of the inauthentic Facebook content we've been seeing; it appears to be coordinated.

#FluTruxKlan #cdnpoli #TonkaTantrum #OttawaOccupation

https://twitter.com/BySteveReilly/status/1491168971152044033?s=20&t=oAsvKyErf8eLh1meZj5ZGg

I'm skeptical about this. Twitter's business model rewards interaction, regardless of type. It remains to be seen whether downvotes will punish trolls or reward them.

https://twitter.com/weeabread/status/1491464846696763400?t=Z9O_fNAi4_CXeH_oBenxnQ&s=19

It depends. When in doubt, look at their tweets/replies. Are they constantly saying the same, wrong things? Do they admit they're wrong when proved so, or do they pivot to another false premise? Do they frequently say "what about (other thing)"?

https://twitter.com/DannyTheZebra/status/1491834607997575175?t=YYE5-a7G4XliYELdwisf3g&s=19

Others in this thread have mentioned "Sealioning", which describes bad-faith actors insisting that they are acting in good faith. Watch for "just asking questions", "honest question" etc., especially a pattern of this type of exhortation.

They absolutely do! Bots/trolls generate interaction, which generates traffic, which causes more ads to be clicked on and seen.

They also artificially inflate the total number of users on Twitter, boosting its apparent value to potential investors.

https://twitter.com/just1momentpls/status/1491840017181790208?t=1uswLzLIvQq-SyBTJ-MhPw&s=19

Didn't read the thread, part 3: mentions me while trying not to give me "credibility", while doing the usual whining about how blocking someone is "censorship".

#FluTruxKlan #FluTruxKlanGoHome #cdnpoli