A Twitter Bot is a computer program that posts tweets on Twitter. Many of you may not be aware of it, but many of Twitter’s posts are created automatically by computer programs. Some are useful but many are spammers looking for better search engine positioning or get rich quick schemes.
Also, earlier this year, there were allegations that a significant portion of Trump’s Twitter followers were actually Bots designed to increase the perception of social media engagement. Additionally, in April, I shared with you the story of Andrés Sepúlveda, who alleged that he helped Enrique Peña Nieto win the 2012 presidency by hacking adversaries. (link) One of the techniques exposed was the use of automated Twitter bots for followers and user engagement.
I had been interested in Twitter Bots for many years but I had not had the time, nor a sufficiently important reason to create one.
The Donald Trump supporters arguing with a computer robot gave me the reason I was looking for. So, this weekend I created a Twitter Bot that posts Twitter messages in response to posts mentioning the Donald Trump Twitter account. It is a crude Bot but it gets the job done.
The Bot works by polling the Twitter feed every 30 minutes for Donald Trump’s Twitter handle. Those are tweets that are directed at Donald Trump’s account, or in response to a tweet from him. From that list, the Bot creates a list of tweets that had been shared the most and creates a text corpus containing all of that content in an endless text string.
I then created a Markov Chain engine that takes the text sample the Bot had compiled and randomly makes a sentence from it. It is the simplest application of artificial intelligence (AI). The cool thing about Markov chains is that its randomness artificially generates some pretty funny sentence constructs. Obviously, there is no intelligence, just random sentence construction. As a result, many of the sentences the Bot creates is gibberish. However, Markov chain bots have historically created some sentences that were somewhat coherent, especially in the e-book realm, and thus some sentence constructs are funny.
That was essentially what the Donald Trump supporters were arguing against, a Markov chain.
In my case, my Bot isn’t as sophisticated as the one that caused people to argue against it because my Bot just takes posts about Trump and generates a Twitter post from them. Because I do not know what it will post, I programmed it to append #nevertrump at the end of each post. My Bot, does not reply to posts directed at it, yet. I may choose to tweak it later when I have some time.
Yesterday, in the afternoon, I finally got the response I was expecting. As unsophisticated as my Bot is, it still managed to get someone to respond to it.
Now I’m thinking that it is time to add the functionality to respond to individuals like the one above. Who knows, someone might go ballistic against my Bot. Hmm, maybe, some canned sarcastic comebacks. What do you all think?
However, my Bot doesn’t actually follow Twitter guidelines so I don’t know how long it will last until Twitter steps in and kills off my Bot. I launched the Bot on Saturday and as of yesterday, it already had four followers, a few favored Tweets and over 100 posts. Oh, and one Trump supporter calling my Bot a dumbass. As I expected, a few of the posts it created are funny, at least to me. One even posted a naughty word.
Check out my No Trompas Bot at Twitter.com/notrompas before it is terminated by Twitter.
Now that I have the rudimentary program, I may decide to launch a few other Bots, probably one focused on Hillary Clinton. The options are endless. I am also open to any suggestions you may have. Remember, the Twitter Bot automatically tweets whatever you want it to at the frequency I program it. It can tweet random posts or we can create an if…then algorithm to create posts in response to certain key words. So if you have any ideas, let me know. In the meantime, enjoy the Bot before it is terminated where it stands.