The inexplicable rise of open floor plans in tech companies

Guessing you're referring to something by Christopher Alexander, Notes on the Synthesis of Form or A Pattern Language? Those predate your timeline by a few decades but I couldn't find anything else that fit. Whatever it is, I don't think it has quite the seminality in software architecture that you imply...

On the whole, architects tend to be a fairly illiterate group and the practice of architecture is largely a craft creating one off designs, so there aren't really any standard references on office layout or design - and certainly not any theories as theories of office space. There's just a guild that where the academy was still debating the utility of CAD in the year 2000 - I know, I was there. It's still not sold on Building information modeling - aka "using a database."

In so far as architecture is a science, actual buildings are its experiments [an idea stolen from E. H. Gombrich] Unlike a program, there's one shot at the artifact - the building gets built and the budget spent and there's no way to do test driven development or serious integration testing, let alone refactoring or versioning. The plans are like code and the construction process is their one and only pass through the compiler.

Anyway, you might find something in an architectural school library on office layout, but in practice a few pages from something like Time Saver Standards for Building Types is about as much reading as a practicing architect is going to do. The big drivers of schema are soft - the psychology of clients and the architect's empathy for those who have to inhabit the space...that's the whole of whether offices in the core and serfs by the windows gets built or not [and that's not the way to bet].

With a specific open office space, bringing in the modular systems vendor is how the actual layout of cubicles is done. The reason is that architecture is commoditized to the point that chasing through endless swaps of Suzy and Joe and Clay among the cubicles [and whose name goes where is an O(n^3) PITA] and only the person actually selling systems furniture can afford the handholding (unless you're Frank Gehrey and explicitly getting paid to do it and able to assign a $15 an hour intern with an March from Harvard's GSA).

If I were inclined to argumentation I'd point out that architecture was a couple of thousand years old when Vitrivuvious dedicated his 10 books to Augustus Caesar and thus you can pretty much be assured that any proposition that links software and architecture is grounded in all that history and floating free upon the mere sixty years that the concept of software has existed.

Habraken's Structure of the Ordinary is the way territoriality and control express themselves spatially and I find it informative for the sort of problems upon whose solution good office design depends. It's not rules for layout but the effects that various configurations create.

http://www.habraken.com/html/structure_of_the_ordinary.htm

Anyway, if there ever was a field of obsolescence ripe for disruption, it would be architecture here and now - and of course it's already happening. The infrastructure needed to provide many services ain't no building any more. The guild thinks technological progress is faster horses to speed up the production of drawings - or to anchor the analogy better pens. The idea of providing better solutions is inconsistent with the model of architecture as craft because a better solution might not require a building and with a business model that couples fees to a percentage of construction cost, there's no money in it.

Gosh, that was fun.

This is pretty much the optimal situation for an openspace floor layout. Now compare that with the situation in a more mature company. The first thing any sane lead engineer wants to do once v1.0 is out the door and making money, is to pay down on the inherited technical debt. Organise code so that it is modularised, so that changes in one area don't have a major impact on other areas. As money is being made, the team starts to expand as well, so simultaneously you have less need for close interaction between team members (due to improved software structure) and greater opportunity to be distracted by irrelevant discussions, as the scope of the product has now grown so much that most of the engineers will never master all of it, but instead become experts in one particular aspect.

Let me give you a real-world example from my current job. At the moment, I'm deep in the source code of an OpenGL driver, identifying performance problems. The thing is, we already have a product, and it works well, but it uses a 2D graphics APIs, and the designers want iOS7-style blurring. So I've re-implemented the graphics APIs using OpenGL. This is completely independant work. No-one else knows OpenGL, and none of them even see the API, as it's hidden behind a facade that they already know. I had pretty much zero interaction with the team whilst implementing the OpenGL backend, as all I had to do was implement an existing API. I now still have zero interaction, because if the team don't know much about using OpenGL, they know even less about how it's implemented under the hood. I don't need to know about their problems with what happens when the user presses "back", and they don't need to know about my problems with the vsync callback.

Yet, we are still in an openspace. The designer with the ridiculously loud laugh is still distracting me every thirty minutes or so. The two guys having a discussion about the latest ministry shake up in the government, yup, they're doing that about 2m from my ears and not making it any easier for me to concentrate. They programmer talking about the first run sequence with a project manager one desk over? Absolutely of no interest to me.

OK, so much for why openspaces objectively suck once your past the project startup phase. Why do ex-developers / current-CEOs still stay with openspaces? You would have to ask them, but maybe it's because they remember how wonderfully productive it was at the start of the adventure and don't realise that the situation has changed. Or maybe their much more interested in raking in the cash,and are now just waiting to flip the company - future productivity is of no interest. Keeping the costs down to make the company more attractive to buyers though, that might be very interesting to them. Anyway, the fact that they persist with openspaces is not necessarily a ringing endorsement for openspaces.