NOTSOF [p.1-11]

Big thanks to Anna in Moscow who recommended a podcast to Victor in Yekaterinburg who recommended it to me when I was still in Dubai. I began with #70 – Christopher Alexander – 1/2 Notes on the Synthesis of Form. I’d read it before, or at least think I might have, as I remember sitting in a tutorial where Chris Hilford, our first year master, was trying to get us excited about it. I don’t remember anything of the discussion and nothing of the book except it had diagrams.

The podcast discussion is still fresh in my mind and so, 56 years after the book was first published in 1964 and 46 years after I maybe read it in 1974, I thought I’d revisit it and try to understand why people thought it so important. I can’t un-know everything that’s happened since so I’ll be reading it as an historical document and product of its time and place.

“In 1954, Alexander was awarded the top open scholarship to Trinity College, Cambridge University in chemistry and physics, and went on to read mathematics. He earned a Bachelor’s degree in Architecture and a Master’s degree in Mathematics. He took his doctorate at Harvard (the first PhD in Architecture ever awarded at Harvard), and was elected a fellow in 1961. During this period he worked at MIT in transportation theory and computer science, and at Harvard in cognition and cognitive studies …”

I’ll stop it there. Alexander’s PhD was the basis for NOTSOF and hands-up I’ve not seen the actual dissertation but I’ll assume it’s the PhD plus some minor revisions. NOTSOF is said to have been required reading for researchers in computer science through the 1960s and, I can attest, for architecture students in the 1970s. But why? Let’s see.

The book’s argument is stated quickly and simply in the first few pages of the introduction which, not wasting any time, is given a title – The Need for Rationality.

	1.	Modern design problems are too complex to be solved by intuitive design methods suited to smaller and less complex problems.
	2.	In any case, the modern world is too modern and fast-moving to wait for vernacular design processes to arrive at solutions incrementally perfected for their time, place and available resources.
	3.	Design problems are like other problems in that they can be broken down into smaller problems that are then solved individually but it is still not possible to do this for large and complex problems.
	4.	Computers are good at solving large and complex problems that have been framed in a way they can solve.
	5.	This book proposes a way of framing those problems.

I’m uneasy about those first two premises because they’re exactly what Patrik Schumacher used to justify parametricism 50 years later in The Autopoiesis of Architecture. I’m not even five pages in yet I feel I know where this book is going. But there are differences. Patrik Schumacher would arrogantly reject any project data not relevant to The Architect’s Project [c.f. The Autopoiesis of Architecure Vol. I] whereas Alexander simply assumes the chosen project parameters are necessary and sufficient. Both claim their approaches to be more rational and efficient compared with those of intuitive designers who do it all in their head. I’m not going to say either approach will produce better results, but only that the two aren’t as different as they’re being set up to appear.

The book’s first sentence is “These notes are about the process of design; the process of inventing physical things which display new physical order, organization, form, in response to function”. The following three sentences tell us how functional problems are becoming more complex all the time and beyond the ability of designers to solve them, and the following eleven go on to talk about conflicts between performance, jointing, simplicity and economy in the design of a simple household object such as a vacuum cleaner. Alexander argues that “The need for simplicity conflicts with the fact that the form will function better if we choose the best material for each purpose separately. But then, on the other hand, functional diversity of materials makes for expensive and expensive joints between components, which is liable to make maintenance less easy.”  This is not necessarily true, as the quest for simplicity of manufacture could equally well result in fewer parts, materials and processes, all of which could conceivably offset other costs and without necessarily affecting performance. I know I know. It’s only an example and what’s the point of examples if not to mislead?

	1.	Modern design problems are too complex to be solved by intuitive design methods suited to smaller and less complex problems.
	2.	In any case, the modern world is too modern and fast-moving to wait for vernacular design processes to arrive at solutions incrementally perfected for their time, place and available resources.
	3.	Design problems are like other problems in that they can be broken down into smaller problems that are then solved individually but it is still not possible to do this for large and complex problems.
	4.	Computers are good at solving large and complex problems that have been framed in a way they can solve.
	5.	This book proposes a way of framing those problems.

Moving on, we’re told the human capacity to solve arithmetic problems is limited unless we break them down into smaller problems and solve them. Plato said as much, in the frontispiece. So there.

However, when the problem becomes too large it quickly becomes beyond us and Alexander says – just like Plato did – that we need a way to describe larger problems in terms of simple ones. Using mathematics to illustrate the difference between simple and complex problems and the human ability to solve them is fair enough and, sure enough, any complex mathematical problem can be reduced to simpler mathematical statements that a computer can solve. Lest we think of mathematics simply as arithmetic, Alexander informs us (on p.6) that mathematics deals with questions of order and relation and not only with questions of magnitude.

For now, we’ll have to take his word on that. I’ll ignore the pattern word for which Alexander would later become known. It’s more important to state that now, as then, we simply don’t know the mechanisms and processes by which the human brain collects and organizes data, and extracts and makes what it feels or deduces are relevant connections between that data – but it does, and it’s called designing. If you want to replicate this process digitally, then you’ll first need to collect whatever information you somehow think are relevant, decide which you somehow think are important to the problem, assign parts of that problem a priority you somehow think is appropriate, connect variables you somehow think are connected and in the way you somehow think they are connected, and assign weightings to those connections according to how important you somehow think they are. This process may be digitized but that doesn’t mean it’s any more appropriate or accurate than the intuitive one.

The conceit is that the analogue and subjective world can be completely and accurately translated into data and sets and relations. This isn’t even possible for something as causal and physical as medicine. [If it were, we’d have no such thing as sickness or disease. People are on the case.]

	1.	Modern design problems are too complex to be solved by intuitive design methods suited to smaller and less complex problems.
	2.	In any case, the modern world is too modern and fast-moving to wait for vernacular design processes to arrive at solutions incrementally perfected for their time, place and available resources.
	3.	Design problems are like other problems in that they can be broken down into smaller problems that are then solved individually but it is still not possible to do this for large and complex problems.
	4.	Computers are good at solving large and complex problems that have been framed in a way they can solve.
	5.	This book proposes a way of framing those problems.

It’s true computers can solve many problems once variables have been converted into values and formulae but there’s no guarantee they’ll be solving the right problem or in an appropriate way, or even that the problem has been correctly identified and perfectly and completely converted into a form that can be speedily and efficiently manipulated by computer. Harvard PhD granters may well have decided to brush this huge niggle under the carpet but, for a book that came out in 1964, nobody’s said anything. [Who were Alexander’s supervisors, his reviewers?] To be fair, the same criticism can be levelled at the more intuitive design processes as both, in their own ways, skew data and manipulate it to achieve a desired result. The pseudo-rational design process creates an aura of objectivity and transparency while the psuedo-intuitive design process creates an aura of mystery and opacity. Maybe later chapters will clarify but, for now, it looks like Alexander simply wants to replace one form of hocus-pocus with another. Then again, this is how architectural theory usually refreshes itself.

Alexander was a mathematician and excited about the potential of computers to solve problems dealing with questions of order and relation and it must have been hip in 1960 to say computers were the future. In the early 1960s computing was an emerging technology showing much promise that, it must be said, has for the most part been fulfilled beyond expectations. A patent had just been filed for magnetic storage media that were to come to be known as disks but punch cards and punch tape were to remain the dominant means of data input and storage for at least another decade. Key punch operator was an occupation. The 1960s being the 1960s, you had to be a man to get to play with the big hardware.

Plato’s message may have been to break complex problems down to simpler ones but Alexander’s was that we need to let computers get on with it. Someone must have seen much promise in Alexander’s PhD because, soon after in 1961, he was duly elected a Harvard fellow. He was 25. He’d definitely created a product. It was still an academic one but potentially much much bigger.

It’s said that if you have a hammer then every problem looks like a nail. You need the nails to show how useful the hammer is. What Alexander did was provide those nails.

	1.	Modern design problems are too complex to be solved by intuitive design methods suited to smaller and less complex problems.
	2.	In any case, the modern world is too modern and fast-moving to wait for vernacular design processes to arrive at solutions incrementally perfected for their time, place and available resources.
	3.	Design problems are like other problems in that they can be broken down into smaller problems that are then solved individually but it is still not possible to do this for large and complex problems.
	4.	Computers are good at solving large and complex problems that have been framed in a way they can solve.
	5.	This book proposes a way of framing those problems.

The notion that design could be computerized was a powerful one with huge commercial implications as it would remove designers and their inexplicable and inefficient design methods from the process of design and construction.

It was the most industrially powerful idea to pass through Harvard since Gropius threw craftspersons under the bus of mass production.

On page two and fifteen sentences into the book is the following passage.

We’re now talking about design as the process of inventing physical things which display new physical order, organization, form, in response to function, but function is no longer about performance, jointing, simplicity and economy, but now encompasses ecology, human desire, social expectations, law & order, and food distribution. Alexander is not only implying that all these are candidates for rational analysis, but that he can extract and reformulate these abstract and subjective qualities as functional relationships. I don’t think this is possible and frankly I’m scared to think anyone thought that even though people still do.

I also thought designing a complete environment for a million people was a strange example of a design problem an architect would ever be asked to solve. And then I remembered Chandigarh – Le Corbusier’s masterplan for a city of a mere half a million people. In 1951, the Corbusier had been appointed to take over the city’s masterplanning and Chandigarh was mostly complete by the early 1960s – a period that perfectly overlaps Alexander’s PhD and the publication of NOTSOF. I’m not suggesting NOTSOF can be read as a critique of Le Corbusier. Without mentioning Le C-word, it is a critique of Le Corbusier or, at the very least, a critique of a design approach claimed to be intuitive and widely believed to be intuitive.

And nor is it a particularly veiled critique for, in a nod to the title of the introduction, the final two pages of the introduction are a critique of design decisions unsupported by data.

The emerging Post-Modernists were to accuse Modernists and their descendants of choreographing people’s movements within space [although Post-Modernists have never really been called to account for choreographing peoples’ emotional responses to it]. Harvard and the US architectural media marketplace were shall we say “receptive” to a critique of “intuitive” and “unscientific” – a.k.a. “European” or “Corbusian” design approaches, and especially for a critique leveraging the US’s computing superiority as the way forward. Computers promised to turn Alexander’s academic product into a commercial one on an industrial scale. In passing, future demand for new towns in India must have looked strong in 1960.

An appendix contains a “full” “worked” example for the complex design problem of a new town in India. I don’t know if or how much of the ideas contained in Notes on the Synthesis of Form were ever implemented in India or anywhere else, either with or without computers.

Computers of course, were here to stay and it’s said (Wikipedia) that Alexander’s book had an influence on programming language design, modular programming, object-oriented programming, software engineering and other design methodologies. It hardly matters now if this is an overstatement or not. Algorithmic design has automated the less glamorous parts of the design process and parametric techniques have automated the more glamorous parts and in both cases the results are sent directly to production with nary a worker or craftsperson in sight.

That was Chapter I. Introduction: The Need for Rationality. We’re up to page 11. This is going to take longer than I thought.

• • • 

A New Formalism

Old Ideas for New Architectures

The Dacha

The Notebooks of Ludwig Kurz