Preface to Chapter 6

Writing this chapter has been mixed up with developing the augmented archive at the.natureof.software. The goal, as I mentioned previously, is to provide a space for the community of subscribers to read the entirety of the work in one place, as well as—eventually—discuss and annotate it, and further distill the concepts into material that can be applied directly, and where applicable, programmatically. This is still a work in progress, with a handful of items outstanding—in no particular order—to reach “minimum viable” status:

• The visual design, which I will undoubtedly noodle on ad infinitum,
• generated / boilerplate content, and the infrastructure to support it,
• some mechanism that syncs subscribers from the newsletter to the website…

…and later on, when I implement them, dynamic elements and structured data. Whenever possible, I will be streaming these interventions on Twitch and YouTube.

The current list of subscribers, as of this writing, will have access to the.natureof.software. through an initial manual sync. To log in, use the same e-mail address where you get the newsletter delivered. I want to underscore that this site is still lumpy as heck and I am actively hacking on it. I'll try to get automatic syncing online as soon as possible.

And now, enjoy Chapter 6: Good Shape.

---

In the chapter in Nature of Order introducing good shape, Christopher Alexander admitted that it was one of the more elusive of the fifteen properties. Not only is it the only one of them whose name explicitly expresses an opinion, he defined it in terms of itself: analogous to strong centers, a good shape is made up of good shapes. He nevertheless got more specific with an enumerated list of criteria for what he meant:

1. High degree of internal symmetries.
2. Bilateral symmetry (almost always).
3. A well-marked center (not necessarily at the geometric middle).
4. The spaces it creates next to it are also positive (positive space).
5. It is very strongly distinct from what surrounds it.
6. It is relatively compact (i.e., not very different in overall outline from something between 1:1 and 1:2 — exceptions may go as high as 1:4, but almost never higher).
7. It has closure, a feeling of being closed and complete.

The Nature of Order, Book 1, Chapter 2.6, p. 183.

For me, the most salient of these ingredients is probably the aspect ratio, because exemplars of good shape rarely contain elements that are excessively long and skinny (that is, more than four times longer along one dimension than the next longest). Other features, like being convex (at least on average) and closed, having a conspicuous centroid, containing lots of symmetries, etc., all follow from what we know of the other properties. If any of the assertions are surprising, it's the one about bilateral symmetry—that there are respective notions of “left” and “right”, and that they are mirror images of each other. (Why it's surprising will have to be a discussion for the the next chapter, Local Symmetries.)

Worth noting that the plane along which bilateral organisms are symmetrical is called the sagittal plane, the etymology of the term being “shaped like an arrowhead”—that is, an object which has reflection symmetry about an axis—another circular definition.

Indeed, good shape is one of those “fundamental” properties that calls the notion of fundamentality into question: a clear and distinct center, not too bent or jagged, not too long and skinny, made up of other elements with the same characteristics. A skeptic would say most of these are already accounted for: too amorphous is not a strong center, too jagged and it will not create positive space or local symmetries, no recursive composition means no levels of scale or gradients. Yet good shape seems to yoke these all together in a way that is meaningful in its own right.

In contrast to the introductory chapter on the subject in The Nature of Order, which shows almost exclusively specimens of good shape, I have chosen mainly to do the opposite. I did this because per usual, Alexander's examples are almost all ancient things. I'd have to try really hard to find specimens of good shape in the contemporaneous built environment—a safari that I can't afford the time for at the moment. So I'll give you one goodly-shaped object from the modern era that managed to pop into my head: the Curta calculator—a humble, practical device invented under the most exigent of circumstances.

“Bad” Shape

A survey of architecture from the modernists onward suggests good shape is the singular property latter-day architects are most allergic to. The kinds of forms that get hailed by critics as “bold” or “dramatic” tend to be in diametric opposition to the concept. In particular:

• Long skinny things everywhere,
• lots of jagged (or otherwise bulbous), asymmetrical shapes,
• cantilevers, cantilevers, cantilevers.

I submit, actually, that what makes critics call these shapes “bold” is precisely because they are a conscious departure from good shape. In the language of architectural critique, that's what “bold” ostensibly means.

One thing we can say about architecture from the 20th century onward is that it is largely capability-driven. To be sure, any building project is continually in tension between what the client considers to be a successful project, and the agenda of the architect to develop and refine technique, self-express, and attract new clients. What changed, beginning in and around the 20th century, was the rapid and compounding injection of new capabilities, in the form of materials, technology, and capital. Put another way, architects have more latitude than ever to pursue their own agendas while still nominally satisfying their clients—including their clients' appetites for ostentation. The result is a lot of buildings that look the way they do, simply because they can.

On the flipside, in the mid-20th century we see the rise of image culture, beginning with easily-circulated full-colour magazines. This shift (argued by people like Benjamin, Berger, McLuhan, etc.) directed attention and real social pressure to produce artifacts that photograph well.

Long Skinny Things

Those awful glass railings that pervade shopping malls, art museums, and institutional buildings are my go-to candidate for the long-and-skinny aspect of because-we-can architecture. I won't hesitate to call these things a hazard. Often enough there isn't even a proper grip attached to them, so they don't function as anything you can comfortably hold or lean on. When this is the case, the exposed corners are crazy dangerous, especially for children, who do a lot of running around, and whose heads and eyes are at that height. And finally, a lot of the time they are mounted such that if you took a run at one, you could probably push straight through it, instantly shattering and sending you careening overboard. So they aren't even terribly effective as barriers either.

One of the forms that became more feasible in the 20th century was the quintessential long, skinny thing: the ever-taller tower. Now, there are practical reasons to want a tower, namely that it is a structure that can be seen from far away, and doesn't require nearly as much land area or material (and therefore, labour) as a mound or pyramid. A tower nevertheless needs more material (and labour, and engineering) than a non-tower, so it follows that the entity with the most resources who considered itself to be important enough to be seen and recognized from afar would be interested in building towers. Historically this role fell to religious institutions (including god-emperors), followed by (secular) governments, then corporations, and now seems to be gravitating, somewhat perversely, toward ultra-luxe condominiums.

I am lukewarm on the narrative that every tower is inherently a phallic symbol—at least as far as that being the goal of its architect—for this rather pedestrian reason: if you were committed to avoiding that association but nevertheless what you were after was height, your next best option is a pyramid (not connotative of anything at all), which would cost some quadratic function more money just in land, and cubic in materials. Plus, most towers tend to get built next to other towers, where they are largely undifferentiated (except nowadays for parametric razzle-dazzle). So if a phallic symbol was what you genuinely wanted, you'd have to do something extra—which people still absolutely do.

I personally find towers to be lazy designs: buildings are volumes which could occupy any configuration in a three-dimensional envelope. Towers, however, are a low-risk proposition for land developers, as they yield a large quantity of near-identical inventory, which is characteristically easy both to build and to sell.

Adorning the long, skinny things (at least the newer ones) are more long and skinny things. It is customary for new residential towers, which rarely depart too far from an ordinary pancake-stack of rectangular floor plans, to have their silhouettes altered, often drastically, by some kind of cladding. This is usually some sort of concrete or metal ornamentation, or ancillary structure like balconies, which is generated parametrically, fabricated in a factory, and trucked in on flatbed. Because it's parametric, every part can be different—with the constraint that it fits (53 by 8½ feet, a ratio of over 6:1) on the truck. The results are, unsurprisingly, “dramatic”.

I'm inclined to call this style of parametric cladding—after Learning From Las Vegas—the duck-orated shed, because it combines Robert Venturi and Denise Scott Brown's two (in)famous archetypes. The building envelope—an otherwise unremarkable “shed”—is so heavily transformed by decoration that it starts to take on more of the traits of a “duck”—a building that is its own ornament.

While—among other things—the diesel engine, assembly lines, container shipping, finite element analysis, and CNC make it possible to make a building that looks pretty much any way you want, it's advances in materials that make it possible—nay, necessary—to go long and skinny. Materials like structural steel, safety glass, and reinforced concrete are heavy, and you're paying for it—in more ways than one—by the pound. There is both economic and engineering pressure to eliminate as much material from these structures as you can get away with.

My nomination for patron saint of long and skinny goes to Santiago Calatrava. His structures remind me of whale skeletons, alien musical instruments, or the sunbleached remains of dead palm trees. Functional? Sometimes. Distinctive? Definitely. Good shape? Alexander would have said no.

Jagged and/or Bulbous

Another hallmark of capability-driven architecture is the use of irregular, asymmetric shapes, either angular or blob-like, or sometimes both at once. If good shape is defined (by Alexander) to consist of those that aren't too long or skinny, exhibit lateral symmetry, and are made up of shapes of similar characteristics, this trend flies in its face.

Irregular, acute triangular shapes are a trademark of Daniel Libeskind. Note that the cladding was originally supposed to be all glass, but concerns arose during construction around too much light and moisture inside, you know, a museum, being too cold in the winter, as well as sunlight reflecting into nearby office buildings, and of course, the street.

Architects, like any designers, are a fastidiously deliberate bunch. Everything they do is on purpose. Not only that, every purpose is articulable: they can tell you precisely why, in infinite detail, a particular feature of their oeuvre is one way instead of another. And yet, with these kinds of arbitrary shapes, the narrative gets really threadbare. There is usually some rationale, like “I wanted to make it look like a crystal/organism/whatever”. So it isn't completely arbitrary, but the story isn't especially satisfying.

The Seattle Public Library is almost a good shape—if only it was symmetrical. Would anybody have noticed, I wonder, if it had zigged, rather than zagged?

While the architect may have a story for the building as a whole, this arbitrariness still pervades the structure. The question we have to ask about some protrusion or other is what does this exact configuration contribute? Not only to the raw, empirical function of the building, but the experience of the people using it? If Rem Koolhaas had made the Seattle Public Library undulate in the opposite direction, would it have made a difference? If Daniel Libeskind had chosen a slightly different orientation for the crystalline tumour emanating from the Royal Ontario Museum in Toronto, would anybody notice? Would it impact anything?

When you consider a volume of space, there are so many possible configurations for matter in that volume, so why that particular one? How does it benefit to be in that exact configuration as opposed to some other one? What were the trade-offs? Does it frame a nice view? Does it do something clever with the azimuth of the sun?

---

Zaha Hadid created buildings reminiscent of consumer electronics, or the kinds of slick, expensive (and typically, barely fit for purpose) designer household items you would expect from companies like Alessi. For all I know, though, her buildings could work great; it's impossible to tell just by looking at them. You'd have to talk to regular users—the people who live and/or work in them every day—and hope they answer you honestly. My question here, as with Libeskind or Koolhaas, is do these extreme shapes make a difference? Do they make the building more effective for its users, or merely just more interesting to look at? Is this the shape that maximizes both interestingness and utility, or is one sacrificed for the other?

Rantilever

I'm going to preface this one because I anticipate some pushback. In this section I am not talking about the full gamut of structures that fall under the most expansive definition of the term cantilever. Rather, I am talking about the kinds of structures that stick out like a diving board.

The amphitheatre at Parque Villa Lobos in São Paulo, Brazil, by Décio Tozzi. I would love to see a cutaway drawing of this; I imagine the inside of it would look something like a tower crane.

A cantilever is an exercise in structural engineering that results in the clever and parsimonious use of very strong (and often expensive) material to create a shape that can't be made any other way. There are plenty of situations where a cantilever (again, in the sense I laid out above) is the only sensible course of action. The context in which I suspect we see them the most, however, is as instruments of looking “bold” and/or “dramatic”. One conjecture for why it has this effect is that we come pre-programmed to bristle at these structures, because the places we encounter wildly-protruding shapes in nature, like rock outcroppings and tree branches, are known—over evolutionary timescales—to be unstable: we feel the danger in our bones.

Frank Lloyd Wright's Fallingwater house is one cantilever extravaganza that I'm pretty confident everybody is familiar with.

Cantilevering makes these precarious shapes at least nominally safe, and predictably, it does so at a cost. For one, you need extra material whose role is just holding the rest of the structure up, that has to be much stronger, and arranged in one of a short list of particular ways. This means it needs more engineering attention than some other structure would. You are likewise permanently committing the structure to a potentially costly maintenance regime. Nowhere, perhaps, is this cost more disporportionate than Fallingwater, which, according to its foundation's annual reports, appears to hover around five to six million dollars a year.

It isn't clear how much of that is purely structural maintenance, because Fallingwater is listed as a single line item under expenses, which would encapsulate things like paying docents and groundskeepers, wholesale merchandise for the gift shop, and insurance. It is also worth noting that building materials and techniques have markedly improved from the ones Wright had access to; a contemporary clone of Fallingwater would not have the problems the original did (and if it did, the architect would have been sued into oblivion). Nevertheless, the (in)famous cantilevers started to sag long before the house was even finished.

Okay, What About Code?

Built objects diverge from good shape when they are too asymmetric, jagged and/or bulbous, or too long and skinny, with exaggerated cantilevers taking the latter—and potentially the rest—to the extreme. Pertaining to software, I can deal with the symmetry part very quickly, because it's a recap of last chapter, and likely the next one: Reflection symmetry in code could correspond to implementing an inverse function for every invertible function, as duals and complements are probably the biggest source of symmetry, at least as far as the code itself is concerned.

As for long and skinny, we'll have to think a little harder. Depending on how you look at it, code is either a one-dimensional object, or an arbitrarily-many-dimensional object. In either case, a concept like aspect ratio doesn't really map. The property good shape, when applied to software, may be the most abstracted one yet. Not-good shape in architecture is teased out and tortured by gravity and tension—bending, stretching, twisting—or if it isn't, it's just tacked on top. Good shape is more suited to compression loads, because how does one even pull on something that has the proportions of an egg? What do you even hold on to?

It turns out you can build quite a bit using almost exclusively compression structures; using little if any tension at all. It's how people have been building for millennia. Everything made out of brick, stone, or earth (wood is generally considered a tensile material), prior to the availability of modern stuff like steel, is a compression structure. It's how Alexander himself preferred to build. Relegating oneself to compression of course limits what can be built—namely not anything long and skinny, and therefore nothing very tall. On the other hand, Gothic cathedrals are compression structures (so is the Roman Colosseum), and those get pretty big.

Beauvais Cathedral (a compression structure and ostensibly good shape, despite most of it never having been built) has collapsed twice in the ~750 years it has been standing—but it is still standing. Its design flaws have since been palliated by supportive (tension) ties.

The severity of a failure in a compression structure is going to be proportional to how much material is piled on top of it. A tensile structure, on the other hand, is likely to have much more potential energy by volume of material. The word tension is also used to refer to voltage, which is a difference in electrical potential. When a high-voltage structure fails, it typically explodes with energy. A more purely kinetic example may be something like a flywheel, which, while rarely especially long or skinny, is nevertheless a specialized piece of equipment that has to be tended and monitored constantly, because it stores a disproportionate amount of energy, and is liable to fly apart and destroy everything in its immediate surroundings.

The gist here is something like:

Type	Strategy	Materials	Mass	Potential Energy	Failure
Good shape	Compression	Naïve	High	Proportional	Proportional
“Not-good” shape	Tension	Sophisticated	Low	High	Catastrophic

Abstraction from physical properties of good shape versus not.

While we can draw some superficial analogy about not making modules too big, or subroutines too long or take too many arguments, I am inclined instead to posit that what I have been somewhat lazily calling “not-good” shape in architecture, exemplified by long and skinny, asymmetric, etc., is a manifestation of specialized structure in general. To the extent that it is load-bearing at all, the load is often extreme.

Preordainment

Why I believe the place to look for an analogy to good shape in software is the general notion of specialized structure, is from looking at biology. Nature seems to prefer to stay out of the long-and-skinny business, but will make exceptions when there is a niche application. Like:

• When it specifically needs the length and/or the skinniness for something (think structures like blood vessels, nerves, intestines, even fingers, but also whole species of animals like snakes, anteaters, octopi),
• when filling a volume with just-plain-huge is too expensive (trees, and the giraffes that snack on them),
• when the long/skinny thing is disposable, or at least cheap to replace (hair, antlers, porcupine quills, pine needles, jellyfish tentacles).

That said, nature is satisficing, so you get plenty of long-and-skinny where it doesn't confer any kind of advantage whatsoever, but also doesn't tax the organism so onerously that it can't survive and procreate. You know, just like architecture. Or software.

Nature, of course, mutates randomly, and most mutations are deleterious. Long and skinny is liable to break off or get caught on something, so whatever function it serves has to be worth it. Or, if it doesn't serve a function, it had best not get in the way. Symmetries are a topic for the next chapter, but we'll note here that they are easier both to produce (at least biologically) as well as comprehend.

Ever seen a peacock fly? Because they can, despite what you'd expect—and they arguably do a better job of it than say, a comparatively unencumbered chicken.

We humans, on the other hand, “mutate” our systems with purpose. If good shape is a stout, compact, symmetrical thing made out of stout, compact, symmetrical things, and a not-good shape is not that, we can choose “good” shapes or shapes that are less so. As I have suggested, a not-good shape is one that is potentially subject to extraneous and uneven forces, incurs disproportionate engineering and maintenance overhead, risks failing catastrophically, and even if it doesn't fail, possibly affects the overall service lifetime of the artifact. Or, it's completely superfluous and could be removed. But we tolerate it, because there is putatively a reason for it. I'm sure we can think of a lot of things in software that are like this.

One way this syndrome manifests is when system designers decide how they're going to solve a problem before they even know what the problem is. This is like an architect deciding that their next building is going to have a massive cantilever before they even get the brief. In software, the analogue is preemptively deciding that you're going to use NoSQL, or microservices, or blockchain, or machine learning, or even some particular programming language or other.

Cleverness

The next, related, aspect, is something I hesitate to call “overengineering” (I consider that to be something else); perhaps “being too clever” is more accurate. You can be clever in the small in ways that make a system cognitively simpler in the large, or you can be clever for cleverness's sake. It's a truism, after all, that code is a story you tell to other people—maybe somebody you don't even know, or maybe your future self—what you're trying to get the computer to do. I say this as somebody who occasionally creates “high-tensile” computational structures—small but intricate things that generate a lot of leverage. If I do it, it's to enact a particular set of concepts or achieve a certain effect. This is a discipline I had to train myself into, because I find there is constant pressure not to “reinvent the wheel”. This phrase is used as a bludgeon at least as often as it is a legitimate criticism, which you can inveigh against any expenditure of effort prior to its results being realized, no matter how useful (or useless) it turns out to be.

When Alexander was bidding on the Mary Rose Museum, his team put a lot of effort into a foundation that would keep the building from sinking into the mud in Portsmouth Harbour, without blowing the £10 million budget. This foundation supported the massive arches of the main gallery that would house the almost-500 year old shipwreck. From the outside, the building was designed to blend in with the surrounding warehouses and drydocks, a fact which didn't escape the derision of the critic writing about the museum that did eventually get built (for almost thrice the cost). He dismissed it as a “strange melange of traditional motifs”, while also lamenting how the freshly built museum—which predictably looks like a cartoon UFO luxury yacht—lacked the “elegance” of the HMS Victory moored right beside it. So not only would Alexander have given him what he ultimately wanted, he completely failed to recognize that Alexander's design couldn't have been anything but cutting-edge—it would have been infeasible to build in any other period.

We've all seen admonitions not to be “clever” with our code, but I think it's a mistake to interpret that with too much zeal. Take, for example, the cleverest elementary construct in computing: monads. Do you ever really need a monad? Of course not. People have been programming for decades without them. You barely even need (at least to make your own) monads in Haskell, where monads come from. Are there, however, situations in which monads would make things simpler? Of course there are, that's the whole reason why they were invented: to arrange structures and processes so you can think about them in principled ways.

It's the intent here, that I think is significant, rather than how far into the weeds you get. People study esoteric branches of mathematics like category theory and homotopy type theory for example, ultimately to make code behave more intuitively—in ways that can be more easily inferred—because they yield systems that can guarantee a certain range of behaviour. Contrast that with cute pointer tricks, cute introspection tricks, “golfing” and hairy one-liners, monster regular expressions, or superfluous metaprogramming.

Most software, like most building, however, is ordinary. It doesn't need exotic materials or techniques. Cleverness is leverage, and leverage always means trade-offs, or at least the assumption of risk. As I argued above, good shape is relatively unlevered by definition.

Arbitrariness, or Moral Hazard

My final consideration, with regard to good shape in software, is arbitrariness: the outcome of a decision that wouldn't make a meaningful difference if it had been decided some other way. Or, it does make a meaningful difference, and a decision has already been made, that—whether expressly deliberate or not—serves some agenda other than that of the user.

I will concede some overlap here with preordainment above, but I view that as more rooted in technical expediency, rather than its relationship to external goals. What connects arbitrariness to moral hazard is putting the user last in line for consideration, if not actively undermining them.

I was initially motivated to volunteer endianness as an example of this phenomenon; at least the makes-no-difference part of it. Upon further reflection, I think it's better as an anti-example. The respective proponents of big- and little-endianness have principled arguments for why their preferences are (at least to them) superior. In practice, the order of the bytes doesn't matter, as long as you—where “you” is the minuscule raw-memory-wrangling subset of the population—know which rule is in play. This is more of a situation where you have (in this case) two options that are virtually indistinguishable, but you have to pick something, because you can't leave it ambiguous.

Big-endian is also called “network-endian”, presumably because the most significant bits would get sent across a network first. One potential benefit, I suppose, is if you had an outage in the middle of a transfer, it would at least be likely that the order of magnitude made it across the wire. Presumably the Intel people have a similarly rational argument for little-endianness. I don't know, however, if the same exists for the middle-endians.

“True” arbitrariness definitely happens frequently (or at least has, historically) at the user interface level. This is especially true in games, where there is more of an expectation to make things “pop”. The symptoms are roughly the same across the board: some interface designer has decided preemptively that some button or other is going to go right there, visual hierarchy and/or conceptual grouping be damned. This can be done in time as well as space, by, for example, forcing the user to carry out a process in a particular sequence (as is done in a “wizard”) when from the perspective of the system, any order will do.

Since it deals with points in continuous three-dimensional space, architecture proper has much more plausible deniability when it comes to sacrificing the user's experience in service of some other goal. Overwhelmingly in architecture, that other goal is mainly to look cool. Compromises to user experience are going to manifest in things like steps that are too long and shallow, bad wayfinding, weird lighting and especially acoustics, unusable voids sardonically juxtaposed against a noticeable lack of storage, and spotty wi-fi. It moreover doesn't help that for any sufficiently prestigious architect, there is typically a coterie of apologists, ready to excuse every stubbed toe and bumped forehead as a user error.

My favourite example of architectural apologia comes from the couple who got a grant from the French government, to restore to its original state, a Le Corbusier condo that had been hacked and slashed by its previous occupants to adapt it into something usable. In an episode of Stewart Brand's How Buildings Learn, they sheepishly admit to having to rearrange their lives in order to use the space.

In software, we seem to have armies of apologists for the entire industry—which I suppose we need, because the excise we impose upon users is so much more pronounced. We might occasionally lie to them and say things like “our site uses cookies to improve performance”, but a lot of the time we just cut to the chase, with “we'd like you to do this for us”. When I think about good shape and how you can't achieve it with long, skinny, irregular, cantilevered, white-elephant structures that do little for the user and a lot for the architect (and sometimes, if they're lucky, the patron), I see a very clear connection between this practice and what we do.

Over the course of his writing, spanning nearly his entire adult life, Christopher Alexander made numerous remarks about the ordinariness of what he was trying to accomplish. Good shape really embodies that. Good shape is legible, quotidian. To call it humble, moreover, would be ascribing too much self-consciousness to it; rather, good shape is un-clever, almost naïve. It isn't angling for an award or a promotion, it isn't contemplating its legacy. In Alexander's own words, it's “just trying to do a good job.”

One fact that is absolutely dead-to-rights about contemporary architecture is that it is painfully self-conscious. (This was a theme going as far back as Notes on the Synthesis of Form.) It is evident that architects have to balance a number of equities so that their buildings get built in the first place, and so they continue to get work. They have to win—and maintain—the confidence of their patrons, and every project is as much an advertisement for their services as it is a verdict on their competence. Sadly, the users—and the larger public—only factor in if something about the building is so acutely bad that one can't help but openly complain about it.

Rafael Viñoly's car-melting death ray building in London springs to mind. I also already nodded to the staircase in Moshe Safdie's Musée des Beaux-Arts. Both of these architects are still getting jobs. Curiously, the much-vaunted Sydney Opera House did kill Jørn Utzon's career, but it did that at least in part because it cost over ten times its budget.

My mind keeps returning to what I can only call “Darwinism in the most boring possible sense”. Namely, that the things that survive and procreate are the things that survive and procreate, and it can get very difficult to tease apart whether it was because of one trait or another, or in spite of it—especially when considering its interactions with the environment over some particular interval. Put another way, a trait that was adaptive (or maladaptive) may no longer be, in some other time or place.

Software development, just as architecture, is in a technological milieu where we can make anything look like anything, and we're in a media milieu that rewards ostentation and punishes homeliness. It is lamentable, but not surprising, that we get vacuities like “boat-related building looks like boat”. Christopher Alexander's buildings are unambiguously homely when compared to his contemporaries, but they attend to the the user in ways that most architecture can't begin to reach. For this—or more specifically, for announcing that doing otherwise is immoral—he earned the disdain of his fellow architects, and foreclosed on many of them adopting his ideas.

We are also situated in a political economy that can't plan past the next quarter—or at best, the next election.

Where we have room to grow, I submit, is the narrative. Good shape, I have argued, is structurally low-leverage, and that is the common thread that connects the concept from buildings to software. My long bet is that ordinary people are going to get tired of bullshit. They are going to increasingly demand straightforward design rationales that are independently verifiable, and affordably so. If you need higher-leverage structure to serve the user, you should be able to just say so. If you need a peacock feather or two in order to sell the thing, you should be able to just say that as well. Only by selling the thing, after all, is anybody ever going to get to use it—and subsequently develop a sense of how your offering is different. If you can fuse those two concerns into the same structure, then fantastic. Your more vacuous competitors will have trouble doing the same.