Starting this semester, this was probably one of the labs I was looking forward to the most, as Twitter bots have a great potential for both humor and literary intrigue. While some use them for purposes of harassment and the propagation of harmful ideologies (which is an unfortunate potentiality for many digital tools), they can be meaningful and worthwhile when used responsibly. For the purposes of this lab, I created a Twitter profile, which can be found here, and I used this tool to create the JSON code for my bot. Here is what that code ended up looking like:
{
"origin": [
"#person1.capitalize# #wish# to #tell# to #person2# the #story# of #storyDesc#.",
"It #began# in #timePeriod# #country#.",
"In #country#, there was a #typeOfPerson#, who, despite being a #adj# #typeOfPerson#, was #adv# #adj#.",
"#person1.capitalize# first #verb# #person2# in #timePeriod# #country#.",
"But, to #person2#, #person1# #adv2# #verb2# a #adj# #typeOfPerson#."
],
"wish": [
"wish",
"desire",
"need",
"want",
"have"
],
"tell": [
"tell",
"show",
"explain",
"describe",
"express"
],
"storyDesc": [
"how #person1# fell in love with #person2#",
"the war between #country# and #country#"
],
"story": [
"story",
"tale",
"narrative",
"account",
"yarn",
"history"
],
"person1": [
"I",
"you",
"we",
"they"
],
"person2": [
"me",
"you",
"him",
"her",
"us",
"them"
],
"country": [
"North #country#",
"South #country#",
"East #country#",
"West #country#",
"America",
"Canada",
"Mexico",
"Brazil",
"Argentina",
"Spain",
"England",
"France",
"Germany",
"Russia",
"China",
"India",
"Japan",
"Korea",
"Australia",
"Egypt",
"Algeria",
"#timePeriod# #country#"
],
"timePeriod": [
"ancient",
"medieval",
"prehistoric",
"future",
"New",
"Neo",
"#ordinal#-century"
],
"ordinal": [
"1st",
"2nd",
"3rd",
"4th",
"5th",
"6th",
"7th",
"8th",
"9th",
"10th",
"11th",
"12th",
"13th",
"14th",
"15th",
"16th",
"17th",
"18th",
"19th",
"20th",
"21st",
"22nd",
"23rd",
"24th",
"25th",
"26th",
"27th",
"28th",
"29th",
"30th"
],
"began": [
"began",
"started",
"gained weight",
"ended",
"came to a conclusion"
],
"typeOfPerson": [
"sage",
"wizard",
"knight",
"warrior",
"professor",
"student",
"man",
"woman",
"lizardman"
],
"adj": [
"vexatious",
"pleasant",
"cheerful",
"morose",
"melancholy",
"phlegmatic",
"sanguine",
"choleric"
],
"adv": [
"rather",
"quite",
"extremely",
"incredibly",
"somewhat"
],
"verb": [
"saw",
"examined",
"glimpsed",
"recognized",
"beheld",
"descried"
],
"adv2": [
"merely",
"simply",
"completely",
"honestly"
],
"verb2": [
"appeared as",
"seemed like",
"gave the impression of",
"manifested as"
]
}
Finally, I used Cheap Bots Done Quick to run the code and post tweets once an hour for about a week. Of the two-hundred-or-so tweets posted as of writing this blog post, here are some of my favorites:
I have set the bot to still post once-per-day, and I plan on letting it continue running for a little while longer, just to see if anything else comes from it. Overall, this was a quite enjoyable process, and I may have to create some more bots in the future, because I foresee a great deal of creative potential arising from the media form.
The Process
When developing my bot, I attempted to create scenarios that one could use to begin a story. Accordingly, some of the potential “origins” include someone beginning to tell a story about how two people fell in love or about a war between two nations, or someone making a short, declarative, but somewhat vague, statement about an event beginning or ending, or someone providing a general description of a person that could launch into a greater tale. My goal with structuring the bot like this was to generate potential story ideas that one could expand on in a fully-fledged narrative. While there are computer programs that can generate substantial narrative content, I do not have the coding experience necessary to even begin understanding how they operate, so I figured I would let the computer start the scenario so that I could finish it later if I so desired.
Thus, on their own, the tweets may not appear literarily valuable, because they only tell partial stories, but I believe that even in their incompleteness, they provide something worthy of merit and consideration. In a way, this Twitter bot reminds me of Italo Calvino’s novel “If on a winter’s night a traveler,” because it presents an anthology of story segments, which, when concatenated together, combine into an object of artistic merit. For me, the value of reading lies not in the completion of a story but in the process itself, as that is where the generation of empathy and the understanding of other perspectives becomes available. While one may not find this readily apparent in the tweets shown above, I believe they demonstrate enough intrigue to allow one to pause, even if momentarily, and consider the world they attempt to depict.
The Result
I am grateful for the web application Tracery, which I linked above, as it allowed me to produce the bot in a relatively easy-to-understand manner. While the JSON code is not too difficult to understand, this project would have taken substantially longer, had I needed to code everything manually. That being said, I did encounter a few glitches with the program, including it not always properly saving the work.
In constructing the bot itself, my main focus was on producing grammatically correct tweets, and I believe that I was successful in this endeavor. However, as a consequence, some of my constructions do not make much logical sense, such as the one which states, “You desire to show to me the story of how we fell in love with me.” This may make sense in the context of a science-fiction story, with doppelgangers or deleted memories, but, as an independent thought, it does sound like it was written by a computer that does not fully understand the process of constructing sensical thoughts in a written format. The limitations of the computer program did produce some humorous content as well, such as the tweet which says, “I first examined him in 13th-century 21st-century Neo South North Spain.” I have no justification for this one, but I still find it quite amusing. Although, I am thankful that the bot did not produce many with this form of illogical syntax.
Tweets like this are a consequence of the coding, where functions have a potential to call on themselves for data, so that there lies a possibility of an infinite recursive loop. While this is generally something to avoid in programming (for obvious reasons), I decided to incorporate this methodology in my bot, so that there would be a latent infinity hidden within a seemingly short program. Moreover, this indicates the potential infinity within digital literature, as never-ending loops pervade the stories we encounter every day, even if they are not always readily apparent. Due to this, I consider my bot to have been successful, because it gives rise to an infinity of the finite, or a “vast Infinitude confined” to steal the phrase of another.
Twitter Bots and Twitter Novels
In comparison to the Twitter novel “The Right Sort” by David Mitchell, I consider Twitter bots to have more artistic merit, because they utilize the limitations of the platform more effectively. While Mitchell’s story was enjoyable to read, and I respect it as a general work of literature, it felt more like he wrote a story that just happened to be published on Twitter instead of in a traditional book, rather than him writing a story specifically in consideration for the structures of the website. While my experience may have been different had I read the novel as it was being serialized throughout a week (since it would have been a part of my update feed, as opposed to its own independent webpage), I feel that there are other Twitter projects which make use of the platform much more effectively. Whether one looks at bots imitating historical figures, or real people writing tweets from the perspective of fictional (or fictionalized) characters, a substantial amount of creative energy is present on Twitter, making a standard narrative seem uninteresting.
Again, I greatly enjoyed Mitchell’s novel, and I respect its value as a work of literature, but it did not seem to fully incorporate the structure of Twitter into its presentation. Bots are frequently nonsensical, as mine demonstrates, but they attempt to present considerable information with limited words and structures, paralleling the limited character-count imposed by Twitter. Both the novel and the bots share the same platform, and must obey the same technological restraints, but while the novel appears to fight against these constraints, the bot willingly accepts them.
Bots and Humanity
This segues into a general juxtaposition between human and computer authors, where our digital world blurs the line between the two. While the bot was written by a human (ostensibly) and all the words it utilizes were at one point typed into the computer by a user, its tweets are still considered to by computer-generated. We have entered into an age where programs can automatically create stories, through machine-learning algorithms, leading one to consider humanity’s role in the humanities in the future. I can confidently say that humans will continue creating stories indefinitely, but there may come a day when computer-generated content is regarded seriously by the general public, because now, unfortunately, relatively few people would consider something like a Twitter bot as anything more than an interesting novelty.
Currently, what computers lack in emotional authenticity, they more than makeup for in being fast and efficient at producing content, even if most of that content is ignored as noisy data, lessening the impact when a truly revolutionary work of programming gets produced. While one could view this as computers mass-producing literature, I would disagree, because the prevalence of such programs does not seem to reduce the value of literary works developed by humans. One day, computer-generated literature may be indistinguishable from texts written by humans, but I believe that most people would still feel an affinity for the human-generated content, as it would be seen as more “valuable.” Whether this evaluation is valid is open for debate, but I think that many humans harbor a latent mistrust of digital technology, even today.
In effect, this Twitter bot does not seem uncanny, because it is fairly obvious that a human did not write it, but I have encountered other programs which have produced that simultaneous sense of familiarity and distrust typical of the “unheimlich” (such as this one). Overall, though, we are still far enough away from producing artificial intelligence, so that the results of even the most advanced programs feel more comedic than artistically valid, and they do not yet fully generate that sense of horror typically effectuated by robots in science-fiction movies. While things may be different in the future, for now, we do not have to fear an uprising of our robot overlords.
Another Digital Humanities Blog 