The Emotional Effect of Creating a Twitter Bot

creating a twitter bot

Creating a Twitter Housemate

Having learned to create his first TwitterBot, Samuel Fry explains how it felt to create a TwitterBot version of his housemate.

A couple of weeks ago I learned how to create a TwitterBot.

I was enjoying a day at The Art of Bots exhibition at Somerset House: a small showcase of art works created by the bot making community. One of the exhibitors was George Buckenham. He was providing a free bot making workshop, inspired by his Cheap Bots, Done Quick! project.

Essentially, Cheap Bots, Done Quick! is a free and extremely accessible bot-making tool. Like many other people visiting the exhibition, I sat in the space and was given a quick run through of how to create a Bot by George. The Bot that I made at the time was called @ArtTechBot. The concept behind him was a simple one: he would tweet fairly banal comments about technology. Sometimes positive and sometimes not.

I created @ArtTechBot in around 30 minutes (including setting up a new email address and Twitter account).

How are TwitterBots created?

I was surprised at how easy it was to create a TwitterBot. They are written in code, but you don’t need any prior coding skills to get started. The Bot that I created, using the “Cheap Bots, Done Quick!” site, was written in Tracery, a generative grammar specified as a JSON (a self-describing form of JavaScript) string. All I had to do was write some basic logic to tweet a number of potential phrases.

I did not over-think the phrases that I wanted @ArtTechBot to say. If you look him up you will see that every 30 minutes he tweets a fairly generic comment about technology, such as: “What’s the big deal with ?”, “Anyone got any fun ideas?” or even “I’m bored of . ZZZZZZzzzzzz”. Roughly speaking, the bot can tweet around 300 combinations of the same set of standard phrases.

What if a TwitterBot was more real?

As I left Somerset House I was still thinking about the potential of TwitterBots.

I should say at this point that I regularly use Twitter and, like many other Twitter users, the presence of TwitterBots has always been an annoyance to me. It’s the same with the #TeamFollowBack hashtag where Twitter users agree to follow each other for the sake of numbers of followers. However, when you know how to build a Bot it becomes very tempting to experiment. With great power comes great responsibility, right?

The experiment that I had in mind was to create a TwitterBot which actually came across as real. Unlike my @ArtTechBot, the tweets should not feel like that have been automatically generated. The TwitterBot should seem like it actually has opinions.

Creating my TwitterBot Housemate

I decided that I needed to create a TwitterBot based on a real personality – or at least a version of one. Who better to base the Bot on than my housemate, Tobias. I told Tobias that I was going to do this. However, it’s hard to prepare anyone for the prospect of having a digital clone.

TobiasBot TwitterBot Housemate

I started by thinking about his interests. He is an Arsenal fan, so the first phrases were all focused on that. “The trouble with Arsenal is they are always trying to walk it in”, was my first thought – the phrase made famous in Graham Linehan’s The IT Crowd, where Moss (the nerdy computer geek) learns the phrase as a way of proving his manliness.

The theme of Arsenal was a good way for me to enable the @TobiasPBot to generate his own thoughts – rather than mine. I used the sentence, “I have to say that {insert player} is my favourite #Arsenal player right now” as a basis. I then replaced the “insert player” text randomly with different current Arsenal players. That way the Bot could decide for himself who his current favourite player is.

This is how I built up the personality of the @TobiasPBot. I would have to create the format of his sentences, but he would decide on the content.

At what point does a Bot become separate from his maker?

When I explain the process of creating a TwitterBot, I think of my role as something like a Dr Frankenstein. I had something in mind for the Bot that I was creating; however, I am as surprised as anyone else what he decides to write. Sure, there are constraints to his writing but the choices are always a suprise.

Who knows what my housemate thinks about the whole thing. Perhaps I’ll ask him at some point. Maybe his Bot will ask him too.



Samuel Fry

Director, Create Hub

Samuel is the Director of Create Hub. He is also a Business Consultant at IBM, working in their Interactive Experience team. Samuel has a history of working with creative, innovative and entrepreneurial companies such as the creative-business incubator Cockpit Arts, in the Creative Economy team at Nesta, with entrepreneur network Virgin Media Pioneers and as the Enterprise Consultant at University of Bristol. All thoughts shared on this site are his own.


All images courtesy of

Author: admin

Share This Post On