Skip to content

Infinite Monkey Architecture

Post date:

The version of the Infinite Monkey Theorem I first heard went “If you give an infinite number of monkeys an infinite number of typewriters then, sooner or later, one of them will type Shakespeare’s Hamlet.” The correct version is “If you give a monkey a typewriter and an infinite amount of time then, sooner or later, it will type Shakespeare’s Hamlet.” This wikipedia article doesn’t debunk the Infinite Monkey Theorem but it does bring some perspective. I paraphrase.

If we ignore, punctuation, spacing and capitals, then a monkey typing letters at random has a 1 in 26 chance of typing the first letter of Hamlet. It has a 1 in 676 (262) chance of correctly typing the second letter, and a 1 in 17,576 (263) of correctly typing the next. And so on. The probability shrinks exponentially. Now, the text of Hamlet contains approximately 130,000 letters and so the probability of typing them all in the correct order shrinks to 1 in 3.4 x 10183,946, or to 1 in 4.4 x 10360,783 if we include punctuation. etc. In other words [numbers], this means that Even if every proton in the observable universe were a monkey hitting keys from the Big Bang until the End Of The Universe, you’d still need 10360,641 protonic monkey universes for a one in a trillion chance of success.

I admire these next people for devising a simulation to find out for themselves what these numbers meant.

They mean nothing, not so much because they’re large but because we can’t imagine a situation where that information would ever be useful. Infinite monkey time is not a good way of getting another Hamlet written but the logical definition of the Infinite Monkey Theorem tells us we can expect one. And we do.

the infinite monkey theorem states that … any sequence of events which has a non-zero probability of happening, at least as long as it hasn’t occurred, will almost certainly eventually occur.

Despite having no practical purpose, the Infinite Monkey Theorem occupies a place in our minds and so must have a reason for it existing. Something about it makes us want to believe that, given enough time, a worthy outcome will surely result. We accept a multiplicity of mindless options as a substitute for creativity despite it being the opposite. It must be a useful idea for someone, because the Infinite Monkey Theorem is always about Hamlet. We’re never reminded that the same monkey will also get around to typing any number of less-renowned and mediocre plays, the shopping list on my refrigerator, the Ten Commandments, and the minutes to our last department meeting. Most letter sequences will be nonsense. Another niggle is that the process involves much wastage of time but this can be fixed by throwing more typewriters and monkeys at the problem. It’s not as if we’re paying them. Even if the outcome isn’t a masterpiece, this idea of having multiple options generated for next to nothing finds fertile soil in the field of architecture.

• • • 

Architectural competitions, whether open or limited, were once a standard way of generating options for a client.

Pitching three options when the job is still up for grabs is standard practice in the field of graphic design. It’s a good system. As a designer, you don’t want to bet everything on the one proposal yet, on the other hand, if you present too many it’ll have the opposite effect of making you look like you don’t know what you’re doing. Three is the right number. You can say good things about each while checking how the client reacts, and then tell them what they want to hear. Much like fortune telling.

Graphic design clients are familiar with the system and expect three options and no more. As explained to me by a graphic designer I once knew, Option 1 is your first thought and what you wanted to do anyway. Option 2 is the opposite, and Option 3 is a mishmash of Option 1 and Option 2. It’s a very simple and effective way of generating three ideas from one. Not knowing what the opposite of your first idea is going to be introduces an element of uncertainty.

A further piece of advice went, “Once you have your three options, make sure Option 2 is the one you want the client to choose.” This goes against any notion of inherent design merit and is just wacky enough to be true. I wish I’d remembered that in 2003 when I was asked to produce three options for a small rooftop extension to a listed house in the UK. I made the mistake of presenting my preferred option third, as if my client were Goldilocks.

I even prepared a table, summarizing the pros and cons of the three options.

Option 3 was my first thought and what I called Option 1 was its opposite in being the most obvious thing to do and easiest to understand. I described it as “simple” but I wanted to imply it was cheapest and looked a bit cheap. Option 2 was the mishmash and combined the low-key approach of Option 1 with the statement approach of Option 3 but its construction and materials made it the most expensive option. I called it a low-key landmark – by which I meant it was pretentious and expensive. My Option 3 was therefore neither cheap and obvious, nor expensive and dubious. Anyway, the job never went ahead so I never proved or disproved the Option Two Theory. Fifteen years later, I thought about it again when I was working in Dubai and my company presented a client with the following three proposals.

At the time, I was shocked to see this page was the only mention of design in all 344 pages of the four-volume design report but I now understand the elegance of this arrangement. If the only thing that matters is that the client pays for it, then all the fancy theory and justifications can be saved for guest lectures, publications, exhibitions and marketing off the back of whichever option is selected. It doesn’t matter which one it is, as much the same things can and will be said. This time, Option 2 was the one chosen to go ahead. And it did – until 2008 of course, when it no longer did.

The three design options were the result of a system known as the in-house or internal competition in which employees work overtime to produce a proposal and have a shot at glory. An in-house jury called The Design Director will identify the two or three proposals he thinks have the best chance of being chosen by the client. This system of generating options works best when the company has already been selected but the design hasn’t.

However, if neither company nor design have been chosen, then the winners of multiple in-house competitions, compete against each other to be selected winner of a wider competition.

This is how the large commercial practices operate. It’s an office paradigm not invented by Rem Koolhaas but monetized by him to great effect, and disseminated with variations by his various protégés who watched and learned how to replicate the system for themselves. This system of architectural production was already up and running in the 1990s when we understood the newly and strangely diverse output of practices such as Herzog de Meuron (and to a lesser extent, Jean Nouvel) in terms of some unifying factor such as careful consideration of the unique characteristics of the program and site and whatever. Gullible me saw the absence of a house style as originality and dedication to the art when all it was, was a more efficient way of developing projects from mutations generated by an intern farm.

Parametricism is an efficient means of generating infinite options because it doesn’t require a team of infinite monkeys and, more importantly, the bananas to incentivize them. Proponents of parametric design believe and claim that a proposal selected from tens of zillions of possibilities must surely be the best. This is because they associate infinite monkeys with creative output, and assume we do too. They omit to mention that the infinite monkeys don’t know what they’re doing. Someone else gets to decide if the output is a convincing substitute for creativity. Compared with deciding that, comparing strings of random letters with everything Shakespeare ever wrote is a doddle.

• • •

I recently bought this book from a lovely independent bookstore in Shanghai called GARDEN BOOKS. It’s been there all of 17 years.

In one essay, Pawley describes an afternoon spent at final year presentations at London’s Architectural Association in the late 1970s. [I’m sorry I can’t fact-check the year as the book is in my office and current pandemic control here means I can’t enter campus for a few days. UPDATE 14/03/2022: I did check it today and was so wrong! The article was from 2001 which was much later than I thought, but still early enough to respond to the newly dominant system of architectural production.] Even in those pre-computer times, Pawley observed that the dominant method of design seemed to be to devise a system that would generate mutations of shape, select one, and then adapt the program to it. The only thing that’s changed in 50 years is the means of generating those mutations.

I’ve said it before. People say that architectural education isn’t responsive to the needs of architecture production but my impression is that it responds all too well. Graduates may not graduate with a wealth of experience but one thing they do know is how to generate options and the narratives [ugh!] to go with them. They graduate perfectly equipped to segue into the world of A-list employment. At least for a while because once they’ve either wised up or burned out, they begin their own practices and perpetuate the system – much like a virus replicates. Perhaps some pandemic control is in order.

  • Contact tracing: Identity the carriers and routes of transmission.
  • Isolation: Limit your architectural exposure to the buildings you either use or see everyday.
  • Vaccination: Form your own opinions.

I’ve nothing against competition or competitions. As with graphic design, a limited choice of options is often good to help a client understand what they want. The 2005~2006 competition for Absolute Towers in Toronto was a competition in which everyone benefited. I suspect the composition of the jury had a lot to do with that.

Competitions and competitiveness are embedded in academia and architectural education. An architecture student’s introduction to infinite monkey architecture begins the first time an instructor asks a student to “play with it” or asks for multiple options of a sketch idea or sketch model. More often than not, these will be options about mass and never about alternative concepts or approaches. And just like the monkey randomly hitting on keys, the option that went before will have no bearing on the current one that will have no bearing on the next one. It’s a pity. Learning in architecture ought to be about learning to limit the universe of possibilities to those that are actually possible. Infinite monkey architecture is the opposite. With infinite monkey architecture, the generators of those random options don’t need an awareness of whether they are good or bad or possible or impossible. Someone else will decide if the output can be made to suit their purpose.


  • says:

    Today was a travel day and I spent my down time reading posts and articles from you going back to 2017. This one was the icing on the cake. Thanks, Graham.

    • says:

      Hello Hugh and thanks! Sometimes posts sit around as drafts for months or years and sometimes they just fall into place in a few days. This one was one of those, so it must have been thoughts that had been sitting around in my mind for months or years. Cheers!