Shakespeare vs infinite monkeys

Strictly speaking, one immortal monkey would be sufficient

- Jorge Luis Borges, The Total Library (1939)

You probably know the story. Start with an over-abundance of typewriting monkeys, add unlimited time, and inevitably they’ll write the complete works of Shakespeare (along with every other conceivable book, nonsense or otherwise).

In today’s post, I’m going to dig a little into this thought experiment, to make the case that the story may be right about randomness, but it’s wrong about both monkeys and typewriters. Along the way, we’ll encounter more than one sticky keyboard, and the UK zoo that actually put this theory to the test.

Armée de singes

The tale of the typing simians has no clear creator, instead it seems to have emerged as knowledge of non-human primates, statistical reasoning, and mechanised writing developed over the nineteenth century. By the early 20th century it was common enough to be referenced in multiple places. For example, it appears in Jose Luis Borges’ 1939 essay The Total Library—itself an echo of a 1904 ‘scientific fairy tale’ by German author Kurd Lasswitz—although Borges refers to the primates typing out not only Shakespeare but all the books in the British Museum.

In between these two we find French mathematician Émile Borel, who in 1913 published his La Mécanique Statistique et Irréversibilité. It contains a vivid image of an ‘armée de singes dactylographes’ slaving away under the eye of watchful supervisors:

Let us imagine that a million monkeys have been trained to strike at random on the keys of a typewriter and that, under the supervision of illiterate foremen, these typing monkeys work ten hours a day with ardor with a million typewriters of various types. The illiterate foremen would collect the blackened sheets and bind them into volumes. And at the end of a year, these volumes would be found to contain exact copies of books of all kinds and languages ​​kept in the richest libraries in the world.

Borel actually puts limits around his Dickensian workhouse scenario, with the monkeys only forced to work ten hours a day, and for a year. 365 million days worth of typing would result, but as we’ll see, that really wouldn’t be anywhere near long enough to come up with a simple children’s book, let alone mirror the contents of the world’s richest libraries.

There are other, even older, ways to imagine an infinite text-generating scheme. In fact Borges refers to one devised by the Roman Cicero around the time of Julius Ceasar’s assassination (in a book, De Natura Deorum, that he dedicated to Brutus, one of those assassins). Cicero was belittling the Epicureans who claimed that the world was made of atoms that could be randomly assembled, without need for a god to intervene. He suggested that those same people may as well believe that throwing countless individual letters of the alphabet on the ground would produce the epic Annals of Ennius.

Whether or not an enormous mess of ancient Scrabble tiles could actually produce Roman poetry isn’t our concern today though. Here I’ll continue with the monkey metaphor. Specifically, I’m wondering what we might learn, by taking that metaphor seriously, about the limits of what we and other animals can produce using technology, and how the form of material objects can affect the paths we take.

Elmo, Gum, Heather, et al.

In mid-2002, Paignton Zoo in Devon, England, installed an iMac all-in-one computer. It was one of those bubble-shaped ones with the semi-transparent blue back that seemed to be everywhere at the turn of the millennium. This one had a few extras, though, including a plexiglass cage, and a plastic keyboard cover that offered a lot more protection than normal.

The precautions were necessary. The iMac was installed in the corner of the zoo’s Sulawesi crested macaque (Macaca nigra) enclosure, and it was part of a mixed experiment/performance art piece. Over the course of a month and a half, the six monkeys in the enclosure had free rein to type any of their thoughts into the computer. A live webcam was set up, and Elmo, Gum, Heather, Holly, Mistletoe and Rowan went to work.

Spoiler alert: the monkeys didn’t write a play, or even a sonnet. Along with smashing a stone on the keyboard, they entered a total of 13,069 characters, and most of what they wrote was an unbroken series of the letter ‘s’ (although the letters g, q and a made decent appearances too). By my count, the longest uninterrupted ‘s’ sequence runs to 7241 characters, over half the text. Let that sink in. Collectively, they pressed that key 7241 times in a row. What’s more, the absence of any vowels other than ‘a’ meant that they didn’t even get around to writing Shakespeare’s name on a title page. (You can read their entire output in a handy book titled Notes on the Complete Works of Shakespeare here.)

The project’s creators knew that they weren’t faithfully setting up a test of the infinite monkey hypothesis. Instead they were interested in the boundaries between artificial and animal life, which were already blurring even two decades ago. As they noted on their (now sadly defunct) project website:

Animals are not machines. Monkeys producing actions are not equivalent to a random generator such as a computer. On the contrary, it is possible that the monkeys will eventually produce the complete works of Shakespeare but not simply because of chance, but also because they can think and learn.

Stretching the facts, we might even say monkeys have already written the works of Shakespeare. After all, there’s only a few million years of evolution between our monkey-like ancestors and literate, sixteenth/seventeenth century bards.

I’ve included a short video made as part of the Paignton project at the end of this post. There are also press reports from the time that seem a little fixated on the fact that the monkeys didn't only use the computer as a typewriter, but also as a socialising area and occasionally as a toilet.

There are two points that I’d like to make about the Paignton experiment. The first is that the monkeys were drawn to only a few keys, predominantly those on the left side of the English keyboard they were given (a, s, q). These were among the easiest to press if the monkeys sat on or near the keyboard facing the wall, as well as those that were furthest from the rest of the enclosure. If you’re a captive monkey just trying to work in peace, that’s the best place to type. Which raises a question: if the iMac was set up on the oposite side of the enclosure, would we see an alternative monotonous output of l, p and k?

Second, there’s no standard punctuation in the monkey output. Even something as seemingly simple as ‘To be, or not to be’ has a comma in the middle. Importantly, punctuation that requires a concurrent press on the shift key is completely absent: a question mark, or the colon in this sentence, for example. They’d struggle to get through even the start of Romeo’s balcony speech for Juliet (‘But, soft! what light through yonder window breaks?’).

Numerous studies have shown that non-human primates have to learn tool use in the wild—it’s not innate—and simultaneous, coordinated use of both hands for a single task is one of the more difficult skills for either monkeys or apes to learn. Is mastering the shift key an overlooked barrier here?

While the Paignton experiment was as much performance as science, then, it does show a few unacknowledged problems with the infinite monkey idea: first, the physical and social environment of the typewriter or computer is likely to affect the output, and second, there is more to successful writing than pressing buttons randomly. No matter how long these macaques persissssssted, we can be confident that they wouldn’t write Shakesssssspeare. They’d trip on the first quessssstion mark.

Fatal attraction

Monkeys aren’t the only primates that find themselves directed towards the same action over and over again. We humans can find ourselves trapped by technologies and associated behaviours that reflect history and circumstance more than good planning. The typing monkey metaphor actually relies on one of the stickiest such contingencies, the form of the typewriter itself.

Or more accurately, the typewriter layout. The keyboard I’m using on my laptop has a very specific layout of keys, beginning with the well-known QWERTY sequence in the top left of the alphabet rows. This arrangement took decades to solidify from the many designs that initially appeared to enable mechanical writing, including round and dome shapes that didn’t quite catch on:

It didn’t have to be like this. There are remnants of a simple A to Z layout even in the modern computer keyboard: starting with the A in the middle row many of the keys are still largely in sequence through D, F, G, H, J, K, L, then they turn for the bottom row and come back with M and N, even ending with Z back underneath the A.

However, things clearly got messy at some point. It’s been suggested that the need for early typewriter salespeople to peck out the word ‘TYPEWRITER’ as part of their demonstrations led to all the letters in that word being put on the same (top) row. And there’s a persistent but probably mythical rumour that the keys were deliberately designed to slow down typists to avoid the machine getting jammed. The problem with that rumour is that the layout was designed before anyone got fast enough to trouble the mechanism, and one of the most common letter combinations—’ER/RE’—is kept together on the QWERTY layout anyway.

Regardless of the precise history, the advent of typing schools and touch typing meant that the layout was encoded into the muscle memory of millions of typists. It became too entrenched to change, even when electronic keyboards came along. More strangely, perhaps, the same layout now appears on touchscreen devices that don’t rely on muscle memory: it’s a culturally sticky configuration. Alternative designs for touchscreen typing do exist (KALQ made a minor splash in 2013, for example), but none have caught on.

This is how we do it

Primatologist Bill McGrew has succinctly defined culture as ‘the way we do things’. The phrase captures the idea that group members do things that are more similar to each other than they are to members of other groups. What the monkeys and the QWERTY keyboard teach us, though, is that there are both physical and social constraints on how things get done. The physical environment may favour typing on one side of a keyboard, while the physical makeup of your brain and hands might make it very difficult to perform two things at once. The infinite monkey question isn’t ‘to be, or not to be’; it’s whether the monkeys can even type a question without its symbolic marker: ?

And on the social side, the things we make, and even the ways we move and talk as a group, can take us down paths that aren’t the most efficient way of getting things done. History always plays a role, cutting channels that those of us in the present and future find difficult to get out of. That channeling history may be genetic, or it may be developmental, or it may just be the need for a typewriter salesperson to quickly tap out the name of their product. Every step a culture takes makes alternative steps less likely, even less thinkable.

Shakespeare wasn’t the only playwright in London at the turn of the seventeenth century. The cultural pathways that led to his particular social environment gave many people access to writing materials, as well as a knowledge of history and human behaviour that underlies most plays. But it is still something close to miraculous that the specific sequences of characters (and punctuation) laid down by Shakespeare exist. Getting another human to reproduce that is impossible in practice, and thankfully unnecessary.

I know it’s not the intent of the metaphor, but if we must study captive monkeys, why not focus on how they live and think and move in their world, not ours? That’s something that not even a million Shakespeares working on a million typewriters can tell us about.

Video. The Paignton Zoo Experiment, 2002:

Sources: Borges, J.L. (1939) The Total Library; https://www.gwern.net/docs/borges/1939-borges-thetotallibrary.pdf || Lasswitz, K. (1904) Die Universalbibliothek; https://mithilareview.com/lasswitz_09_17/ || Borel, E. (1913) La mécanique statique et l’irréversibilité. Journal de Physique Théorique et Appliquée 3:189-196; https://hal.archives-ouvertes.fr/jpa-00241832/document || Cicero (ca. 45BC) De Natura Deorum. || Schuppli, C. et al. (2021) The ontogeny of exploratory object manipulation behaviour in wild orangutans. Evolutionary Human Sciences 3:e39. || Singh, V. (2020) How the QWERTY keyboard shapes the way we communicate. British Academy; https://www.thebritishacademy.ac.uk/blog/summer-showcase-2020-how-qwerty-keyboard-shapes-way-we-communicate/

Image credits: Notes Towards the Complete Works of Shakespeare; http://www.vivaria.net/experiments/notes/documentation/ (accessed via the Internet Archive, July 2021) || Institute of Digital Art and Technology (i-DAT), University of Plymouth; https://i-dat.org/2002-generator/ || Singh (2020)