Thread by @conspirator0

Here's a thread about bots: what bots are and (some of) what bots do, and also things that aren't bots but frequently get mistaken for them.

cc: @ZellaQuixote

First, what is a bot? The Oxford English Dictionary defines "bot" as "an autonomous program on the internet... that can interact with systems or users". A Twitter bot is simply an automated Twitter account (operated by a piece of computer software rather than a human).

Although much public discussion of "bots" centers on malicious or spammy accounts, there are plenty of legitimate uses of automation. Many news outlets use automation tools to automatically share their articles and videos on Twitter, for example.

There are a variety of fun and useful Twitter bots that freely disclose that they're automated. Some examples:

@everycolorbot - tweets hex values of random colors
@archillect - tweets images
@earthquakeBot - tweets earthquake info
@BidenInsultBot - insults you if you tag it

Although forbidden by Twitter's automation rules (help.twitter.com/en/rules-and-p…), spam networks (large groups of accounts operated by a single entity tweeting the same stuff) are a frequent use of automation.

What do spambots spam? Cryptocurrency has been a hot topic for automated spam networks in recent months, with networks ranging in size from a few dozen accounts to tens of thousands. Some examples:

Porn is (unsurprisingly, since it's the Internet) another frequent use for spambot networks:

Spammy retweet botnets are also sometimes used to astroturf political topics. A few examples:

Xinjiang human rights abuse denialism:
2021 Ecuador election:
Random US Congressional tweets:

Sometimes automated spam exists alongside organic activity on same group of accounts. An example of this is the now-defunct Power10 automation tool, which caused its users to automatically retweet large numbers of pro-Trump tweets.
businessinsider.com/power10-activi…

Services that sell retweets, likes, and follows (all of which are TOS violations) frequently use botnets to provide the aforementions retweets, likes, and follows. A couple examples:

The above is by no means a comprehensive survey of every bot or every type of bot on Twitter, but is a decent rough overview of common uses of automation, both legitimate and illicit.

Onward to the next topic: types of accounts that people think are bots, but aren't.

Folks who participate in retweet rooms (where everyone retweets every tweet shared in the room) often get mistaken for bots due to their high tweet volume, as users who are in multiple rooms often retweet hundreds of tweets a day.
politico.eu/article/twitte…

Copypastas (cases where real humans copy and paste the same of text with few or no alterations) frequently get mistaken for bot activity, as identical tweets appear on multiple accounts.

In a similar vein, accounts that share a lot of news articles or YouTube videos by using the "Share" buttons on the respective sites get mistaken for bots because the article/video title is generally used as the tweet text, resulting in identical (but not automated) tweets.

Impostor accounts and fake personas also frequently get erroneously referred to as "bots", although many of these are human-operated rather than automated.
buzzfeednews.com/article/craigs…

How does one know if a given account is a bot? Unfortunately, there's no quick way to tell, and in many cases without finding a large number of accounts that belong to the same network, it may be impossible to be certain. Here are a couple of things that sometimes work...

Every tweet is labeled with the software it was tweeted with, which can be used to identify automated tweets. Most human tweets are sent with Twitter Web App, Twitter for iPhone/iPad/Android, or TweetDeck and most tweets sent with other apps are automated.

Anomalies in timing can sometimes indicate automation as well: constant activity without breaks for sleep, for example. The accounts described in the linked thread are examples (and have other timing anomalies as well).

Every rule has exceptions, of course:

• some 24/7 accounts are run by multiple people rather than being automated
• web browsers and phones can be automated, so some accounts that post via web/smartphone are actually bots
• etc

(just to clarify - the articles linked in this tweet aren't erroneously claiming that things that the fake accounts described are bots, they're accurate articles about fake/impostor accounts that sometimes have been mistaken for bots.)

Also, be wary of overly simplistic checklists that purport to be advice on "bot detection". Most of the stuff on this list has nothing to do with automation, and will be of little to no use in determining whether or not a given account is a bot.

Recent Twitter blog post that covers various misconceptions regarding bots:
blog.twitter.com/common-thread/…