Bots

How it works

A bot is a social media account run by software instead of a person. Somebody writes a script, points it at a few hundred accounts and tells them what to do: like this post, repost that one, reply with one of these twelve sentences, use this hashtag between 7 and 9 pm. Nobody is reading your replies. Nobody is being persuaded. The point is not conversation but numbers.

Numbers are the product. Platforms rank content by engagement, people judge content by engagement, and journalists write “X is trending” stories about engagement. A few hundred automated accounts can push a hashtag onto a trending list in the quiet hours of the morning, where real people then find it and keep it alive. The bots light the match; the humans bring the fuel.

What bots are bad at is everything else. They cannot argue, adapt or sound like your cousin. Pew Research found in 2018 that around two-thirds of tweeted links to popular websites came from suspected automated accounts, but most of that is boring: news aggregators, weather feeds, spam. The political bot is a small, loud slice. Which is why the lazy response, “you’re a bot” to anyone who disagrees, is almost always wrong, and is itself a trick: it lets you dismiss a person without answering them.

Detection tools exist (Botometer-style scores), but they guess from behaviour. They mislabel enthusiastic humans as bots and miss well-built bots. Treat a score as a hint, never as proof. Bots also rarely work alone: a real campaign mixes automation with hand-run sockpuppets and paid people (see troll farms).

Example

@lena_r_1994, 2h The new tax law is a disgrace. #ResignNow

@freedom_voice_4471, 2h The new tax law is a disgrace. #ResignNow

@TruthMatters2026, 2h The new tax law is a disgrace. #ResignNow Wake up people

@gov_defender, 1h @marta.kovacs you’ve criticised the minister three times this week. Obvious bot. Blocked.

Notice two things. The first three accounts posted identical text in the same minute. They were created in the same month, have numbers in their names and reply to no one. That is a network. The fourth is a real person using “bot” as a way to not answer a real critic. Both are worth spotting.

Why it works on you

• You count before you read. A thousand reposts feels like a thousand opinions. It might be one laptop. See bandwagon effect.
• Trends feel like news. When something trends, you assume a lot of people care. The platform never checked who those people are.
• “Bot” is a comfortable label. It is easier to believe the other side is machines than that real people disagree with you. That comfort is exactly what the dismissal sells (see ad hominem).

What to do

When you suspect a bot: open the profile. Creation date, follower ratio, posts per day, whether it ever replies like a human, whether its photo turns up elsewhere (reverse image search). Three red flags together mean something; one means nothing.

When you catch yourself saying “bot”: ask whether you mean “automated” or “I don’t like this”. If it is the second, answer the argument or walk away. Do not hide behind the word.

In a comment section: do not fight the bots; nobody is home. Report the network if the pattern is obvious, screenshot the identical posts, and tell your friends why a trend is not a vote.

Don’t confuse it with

• Useful bots post weather, earthquake alerts or a newspaper’s new articles. They are labelled, they are dull, and nobody pretends they are people.
• Enthusiastic humans. Someone who posts eighty times a day about their team is a fan, not a script. Volume alone proves nothing.
• Troll farms are paid people; they can argue back. Bots cannot.