In this section Koolhaas is talking about being assigned to assess whether an old Jeremy Bentham-inspired 19th century prison--the Koepel--aka the Arnhem panopticon--was still usable considering modern-day attitudes toward incarceration and rehabilitation:
Perhaps the most important and least recognized difference between traditional  and contemporary architecture is revealed in the way a hypermonumental, space-wasting building like the Arnhem panopticon proves flexible, while modern architecture is based on a deterministic coincidence between form and program, its purpose no longer an abstraction like »moral improvement« but a literal inventory of all the details of daily life.
Flexibility is not the exhaustive anticipation of all possible changes. Most changes are unpredictable. Bentham could never have imagined the present use of the Koepel. Flexibility is the creation of margin – excess capacity that enables different and even opposite interpretations an uses. Because Bentham's ideological purity could only be realized at the cost of a spacial surplus (Zak's note: the idea was a big golfball with cells on the rim and guards in the center with lots of empty space in the middle), the Koepel is such a margin. New architecture, lacking this kind of excess, is doomed to a permanent state of alteration if it is to adjust to even minor ideological or practical changes…The history of prison building has become a sequence of short-lived ideals that were challenged, faltered, and then failed. Near the end of the 20th century, this sequence becomes almost comic – like an accelerated movie. It has become impossible to build a prison that is not, at the moment of its completion, out-of-date.
|Lots of room...|
|...for other things|
Basically, the old building had extra space in it. It was an inefficient design, and that's why it's proved useful for over 130 years. That unused space translated into flexibility.
Aside from being a useful parable about how the ideas attending a thing's creation aren't necessarily the only ones that thing can express, it's also a story about durable design. Anyone familiar with the old rhetoric of focused game design ("The game is about what the rules are about") can see where this is headed.
The classic counter-example is the incredibly shambolic and old-school Call of Cthulhu, a perfect storm of traditional-game design Shouldn't-Works:
-Clunky: Up until very recently it was still basically just Chaosium Basic Role-Playing (D&Dish stats + Plus a couple more + Plus a bolted-on skill percentile skill system including some hybrid shoot skill plus dodge chance combat system) + Percentile Sanity Stat + Percentile Cthulhu mythos stat
-Wasteful: Plus 90% of the 200+ page huge hardcover book is stuff you never use in any given session.
-Empty center: Plus the insanities (the defining feature of the game) are on a barely-described chart that take up a quarter-page.
-Terrible examples: The published adventures are notorious for being railroady.
Focused design wonks have no idea what to do with this game, which has been successful in a wide variety of campaigns for most of the history of RPGing and has barely bee altered edition-to-edition.
Yesterday I ran a Call of Cthulhu game that, instead of being about neuraesthenic aesthetes confronting eldritch library horror was instead about a crew of modern cops on the edge trying to clean up a city that likes being dirty. Here is a list of the changes I made to the way we usually run things:
1. Players have to choose the 'Police' career
2. That's all
|Lots of room|
This game has Margin. And it's not like I had to write up things either, like I would with modern supposed-to-be-generic systems. Because the core of the game is simply "What describes things that exist in this place and time?" rather than "What is genre-appropriate?" Cthulhu characters exist in a world that still has a lot of things that, thematically, aren't horror-specific like cars and botany. And it grasps that you'd like to be able to play lots of sessions--either lots of different kinds of games, or lots of the same campaign but with evolving roles and situations, so it needs to weigh things players can do close to equally. Other trad systems think the same way--D&D, Warhammer, even FASERIP can all drift away from their default modes with barely an effort.
Just as no prison can keep up with the state-of-the-art in locking people up, no game can (or should) be able to anticipate exactly what a random one-billionth of the world's population wants to do around a table on any given Sunday.
Next time you ask "Hey what system would you use to run a game of roller-skating magic girls in the 94th century who do combat by proton-strangulation?" and someone says "D&D"--that's why. It's not because they don't know any better, it's because it'll work.
And, unlike Bentham's prison: it was designed that way on purpose.