Oddly Influenced

Welcome to oddly influenced, a podcast about how people have applied ideas from *outside* software *to* software. Episode 18: /Seeing Like a State/, part two: understanding your High Modernist eidolon.

Remember, Scott’s subtitle is “How certain schemes to improve the human condition have failed”. If you look at the failing schemes, it seems incomprehensible that anyone could have considered them likely to succeed. As a minor starting example, the architect and city planner Le Corbusier took over the planning of a new city. He replaced the housing planned for the center of the city with 220 acres of monuments at quote “a great distance from the nearest residences”. This is bizarre. He placed static monuments at the center of the city – typically the place of most importance – and then make it awkward for people to get to them. It was as if the monuments were to exist alone, in solitary glory.

Something was *wrong* with Le Corbusier. What?

Most of us are not Le Corbusier, not Grand Designers Who Make Sweeping Plans, but plenty of people at our level - the level of those implementing grand plans – behave weirdly too. Scott has example after example of minor officials getting obsessed with, for instance, whether villagers had planted crops in tidy enough rows or whether new houses were placed exactly on the radial lines of the village plan. If not – if the houses were as little as 6 meters off center – such officials had them disassembled and ordered the owners to rebuild them right.

What was *wrong* with those officials?

I’ve seen code reviews that were as silly as forced relocation of houses. Somewhere around 2010, I was consulting for a C++ shop that produced code to run on embedded systems. There was a senior engineer who would reject code that used postincrement (that is, variable++) instead of preincrement (++variable) in cases where either would have been correct. Why? Because preincrement was quote “more efficient”. Now, I understood where he was coming from. I wrote C code for the PDP-11 processor in the early 1980s. On the PDP-11, preincrement really *was* faster than postincrement because there was a special opcode for it. But, weirdly, things had changed in almost 30 years. Compilers in 2010 were smart enough to optimize postincrements to match the hardware. But our senior guy was insistent.

What was *wrong* with him?

In 1981, I attended the World Science Fiction Convention in Chicago, USA. I volunteered to be an usher for the climactic Awards Banquet. We ushers were told that no weapons would be allowed in the hall. Then, as now, people liked to dress up as barbarians or whatever.

Someone came up to the door and he had something perhaps marginally a weapon – a stylized cane or something. I think he was dressed fancy.

I told him he couldn’t come into the hall with the “weapon”. I was wrong: it was silly to think his adornment would count as a weapon. He made a fuss, I think, and I was overruled.

What was *wrong* with me?

I don’t know if it was right then or later, but I still remember the profound feeling of “what on earth got into me?”

When writing the script for this episode, I decided to lean into that question. So, in keeping with the science fiction theme I just introduced, I’m going to pretend what had “gotten into me” was one of the *eidolons* that frequently pop up in Paul McAuley’s work. An eidolon is a snippet of alien algorithm that can run on someone’s brain. In most infections, the mismatch between the eidolon and the substrate it’s running on causes strange and compulsive – but goal-directed – behavior. In many cases, the eidolons are quiescent until something in the environment – the context – triggers them. Then they keep their human doing the strange and obsessive thing until the stimulus stops.

Let’s say, just for this episode, that Le Corbusier, those village inspectors, that senior developer, and I in Chicago were all infected by High Modernist eidolons. Seems silly, but I’m making two admittedly wild gambles by using this metaphor.

First, I’m hoping that the image is striking enough that it lingers in your brain, ready to unconsciously trigger a realization that what you’re doing *right now* fits a pattern that’s known to fail.

Second, if you’re like most people, your good intentions make you less likely to accept any claim you’re producing bad results. But if it’s not *you* that’s producing the bad results, but an alien eidolon, well, maybe you won’t talk yourself out of understanding you’re on the road to failure.

And, if nothing else, maybe I’ve alerted you to some fiction you might like to read. See the show notes.

I’m going to say the High Modernist eidolon is triggered by the potential of attaining – take a deep breath here – most or all of efficiency, order, power, simplicity, correctness, universality, and distance. Almost all of those need special modifiers – that is, it’s not exactly ‘efficiency’ in the broad sense – and they’re all mushed up together, so it does no good to treat them separately. Instead I’ll use examples to try to tease out the meanings. I hope this works.

Le Corbusier and Lenin are Scott’s two main antiheros. It’s interesting that both of them were big fans of Taylorism. Taylorism, also known as Scientific Management, was a movement that started around 140 years ago. It was named after Frederick Winslow Taylor, who is famous for measuring factory workers with a stopwatch to discover the single most efficient way to perform a task, and then training workers to make the optimal movements at the optimal speed, for optimal efficiency. He wasn’t, probably, as much of a villain as he’s often made out to be, but he had a certain mania for efficiency that apparently struck a chord with our two main characters.

As an example, Le Corbusier once proposed that factory workers and their families should live in barracks stretched alongside major transportation arteries. Now, Le Corbusier wasn’t a monster: he actually spent a lot of time thinking about how much fresh air, cubic meters of living space, and outdoor recreation area the, um, unit citizen needed. But he’d simplified his problem space down into three elements: workplace, home-place, and transportation between the two. In that simplified problem domain, efficiency of movement between work and home dominated his attention.

He was apparently unable to understand there are *reasons* people don’t flock to live next to a highway, that an empty playground next to a freeway doesn’t really count as “recreational space”, that maximizing the efficiency of travel between work and home is not in fact the most important part of deciding where to live. He wasn’t a dummy, I don’t think. He was just captured by a compulsion that made him think he knew everything he needed to know.

I’m sure it’s historically inaccurate, but I’m imagining a young Le Corbusier who realizes that design is hard. Designing a single part of a city that’s for living, shopping, *and* entertaining involves juggling many more variables than designing three separate spaces, each with a single function. Le Corbusier writes, quote “an infinity of combinations is possible when innumerable and diverse elements are brought together. But the human mind loses itself and becomes fatigued by such a labyrinth of possibilities. Control becomes impossible. The spiritual failure that must result is disheartening.”

So he begins to insist on functional separation for practical reasons. But, ever so subtly, his eidolon twists his mind into thinking that functional separation isn’t a *tool*, but rather a *goal*. Separation doesn’t just simplify, it *must* be present in any correct design.

Under the alien influence, Le Corbusier comes to *hate* mixed-use neighborhoods, like the kind where there are shops on the ground floor, dwellings above, and people who spend a lot of time hanging out on the corners or the streets or in front of cafes. To him, “a bearable, acceptable framework for [man’s] existence, one productive of human well-being and control, [requires projecting] the laws of nature into a system that is a manifestation of the human spirit itself: geometry.” Geometry is simple, and powerful, and it expresses universal truths. Therefore neighborhoods must be geometrical. They must have straight lines, right angles, and a grid layout to be correct.

I think there’s a certain thrill in *imposing* simplicity on a world that doesn’t naturally welcome it. Scott gives examples of people, during the first era of massive public works like the Suez canal, who planned roads or railroads that went in straight lines: up and down hills rather than winding around them. Or, in some cases, they demanded that roads be not only straight but level, no matter how much material had to be moved from the high places or added to the low places. This is certainly not efficient in terms of cost, but the straight line is *such* a universal form, such a *simple* form, and so much more orderly than curves, that the aesthetics took precedence.

We in software, like Le Corbusier, are fans of functional separation, and I for one think it’s most often true that functional separation is necessary for maintainable systems. But what we do – and I think we should *not* do – is turn functional separation into a fetish. Functional separation is not inherently good; it is rare in evolved creatures. Even your heart is not *just* a pump. Functional separation is a simplification we’re forced into. Implementing it is a cause for humility, not pride.

A tool, not a goal. A tool we might sometimes *not* use, even if it makes us feel as creepy as Le Corbusier looking at people sitting outside a sidewalk cafe enjoying themselves.

Scott’s book has a number of boring examples of power in it. There’s the base desire to boss people around – certainly a part of my Worldcon behavior – and there’s bureaucrats jockeying for promotions or influence, but there are also interesting variants.

One is the thrill of new capabilities, new kinds of intellectual control. Here are a couple of interesting, non-Scott, quotes:

The first is from Gombrich’s /The Story of Art/, 16th edition, on the new technology of perspective:

“It was said of Uccello that the discovery of perspective had so impressed him that he spent nights and days drawing objects in foreshortening, and setting himself ever new problems. His fellow artists used to tell that he was so engrossed in these studies that he would hardly look up when his wife called him to go to bed.”

The compulsive programmer, Renaissance edition.

And from Feyerabend’s /Conquest of Abundance/, on logic in ancient Greece as described by Gershenson and Greenberg:

“Logic was the rage of the day; all over, at the marketplaces, in the streets, in private homes and in public buildings, at all times, sometimes all through the night, people engaged in dialectical disputations and flocked to hear the acknowledged masters of logical argument display their art.”

That’s how a charismatic person like Le Corbusier can enroll an entire profession into his obsessions, lasting for decades. I think the appeal of a new power has something to do with why software development is so fad-driven. Each new thing: the Spotify Model, TDD, structured programming, object orientation, immutable data structures, and so on just sweeps some people away until they find themselves behaving like a road planner sending a road straight up the side of a plateau, switchbacks be damned.

Another aspect of power is that it’s not human power: it’s the power of *correctness*. The designer channels the power of an *idea*. Here’s Le Corbusier again:

“The despot is not a man. It is the *Plan*. *The correct, realistic, exact plan*, the one that will provide your solution once the problem has been posited clearly, in its entirety, in its indispensable harmony. *This plan has been drawn up well away from the frenzy in the mayor’s office or the town hall, from the cries of the electorate or the laments of society’s victims.* It has been drawn up by serene and lucid minds. It has taken account of nothing but human truths. It has ignored all current regulations, all existing usages and channels.”

Please note that the shouty parts were emphasized in the original.

Lenin, of course, had absolute confidence in the truth of Scientific Marxism. He knew he could be the conduit for the inevitable progress of history. Not that he’s alone in such confidence. Here’s George Stigler, Nobel Prize winner and leader of the Chicago school of libertarian-ish economics, very much not politically anything like Lenin:

"Affairs of science and intellectual life generally are not to be conducted on democratic procedures; one cannot establish a mathematical theorem by a vote, even a vote of mathematicians. Therefore an elite must emerge and instill higher standards than the public or the profession instinctively desire. The best economics in the US is not the one the public would elect. A science must impose the standards of an elite upon a profession."

Let’s just say that an attitude like that has a poor track record, whether the elite is the University of Chicago, or the Politburo, or programmers deciding how other people should do their jobs.

We seek order, but it’s important to note that order can be either simple or complicated. However, the eidolon is only attracted to simple - typically simplistic - order. An example is scientific agriculture’s aversion to “polycropping”, that is, growing multiple species on the same land. For example, in Tanzania, coffee was typically planted in the same place as bananas, beans, and other annuals. If you’re used to thinking of orderly agriculture as requiring just one species, planted in straight, properly-spaced rows, that just looks like a mess. However, Scott quotes Howard Jones, who wrote in 1936 about Nigerian polycropping:

“To the European the whole scheme seems … laughable and ridiculous, and in the end he would probably conclude that it is merely foolish to crowd different plants together in this childish way so that they may choke one another. Yet if one looks at it more closely there seems a reason for everything. The plants are not growing at random, but have been planted at proper distances on hillocks of soil arranged in such a way that when rain falls it does not waterlog the plants, nor does it pour off the surface and wash away the fine soil… The soil is always occupied and is neither dried up by the sun nor leached out by the rain, as it would be if it were left bare… This is but one of many examples that might be given that should warn us to be very cautious and thorough before we pass judgment upon native agriculture. The whole method of farming and outlook of the farmer are so entirely new to us that we are strongly tempted to call it foolish.”

Someone under the compulsion of simplicity and novel forms of power can’t notice deeper order, especially the deeper orders underlying human behavior.

It’s also worth noting the role of universality. The people who “advised” people in the tropics how to farm better had experience with temperate zone agriculture. Temperate zone agriculture is well suited to monocropping and orderly rows. The soil and growing seasons in the tropics are radically different, but consultants could not see that their rules were derived in a particular context and only worked there.

Scott says, “High-modernist plans tend to ‘travel’ as an abbreviated visual image of efficiency that is less a scientific proposition to be tested than a quasi-religious faith in a visual sign or representation of order.” He’s talking about, for example, village designers in Tanzania who insisted on rectangular plots no matter what the terrain.

We in software have some of that, I think. I personally think the fascination with a single coding standard, preferably enforced by a tool, is more about the aesthetics of visual order than about the practical benefit of utmost consistency. But I’m biased, because I’ve yet to see a code standard that tolerates my strong preference for tabular unit tests. But I’ll leave my argument to the show notes, and I know it’ll convince only those who already agree with me.

More important, though, is that we’re adept at symbol manipulation, so we extend the compulsion for order and simplicity and universality to symbolic structures. We embrace the laws governing abstract data types or Haskell type classes, but we turn up our noses at “business logic” – it’s so illogical. It’s full of contingency – things that are done that way just because they were done that way in the past. Remember from last episode that High Modernists resent the constraints of the past and are not so fond of the details of the present either. We would rather get an abstract description of the problem and reason our way to the solution from first principles and our existing, universally applicable toolkit.

Like Le Corbusier, many of us long to “stay well away from the frenzy in the mayor’s office or the town hall, from the cries of the electorate or the laments of society’s victims”. To be honest, people really can be awfully annoying. “Oh, Mr. social network provider, why won’t you be ethical and think about women fleeing their abusive husband” and so on. It’s enough to drive a formalist to distraction. As Monsieur L.C. says, “the human mind loses itself and becomes fatigued by such a labyrinth of possibilities.”

A good example of the attractiveness of distancing and universality is the Verblud farm in the Soviet Union.

In the early 20th century, people like Lenin were greatly impressed by US scientific farming. Specifically, they were impressed by the biggest, most factory-like attempts at it, even though those had been generally less productive than smaller family farms. US farming consultants saw the Soviet Union as an ideal place to try out their most cutting-edge ideas, ones that were too big or too expensive for the US.

So it was a match made in heaven. Three people were hired by the Soviet Union to plan a wheat farm that would be created from scratch on 500,000 acres of virgin land. “They planned the entire farm layout, labor force, machinery needs, crop rotations, and lockstep work schedule in a Chicago hotel room in two weeks in December 1928.”

I’m gonna say that at *best* they thought their distance from the actual land was irrelevant. I imagine it more likely that they thought it a virtue. Someone else had collected all the necessary information that described the problem, and the distance allowed them to do their work with Le Corbusier’s “serene and lucid minds”. Except I bet they weren’t serene. I bet they were pumped up, caffeinated, with the feeling they were crushing it (or whatever the equivalent 1928 slang was). That’s how eidolons *do you*.

How did the Chicago planning work out? Scott quotes someone named Fitzgerald, “Even in the U.S., those plans would have been optimistic, actually, because they were based on an unrealistic idealization of nature and human behavior. And insofar as the plans represented what the Americans would do if they had millions of acres of flat land, lots of laborers, and a government commitment to spare no expense in meeting production goals, the plans were designed for an abstract, theoretical kind of place. This agricultural place did not correspond to America, Russia, or any other actual location.”

The Verblud farm was a failure.

So, now I hope I’ve put you in the frame of mind of Tony, one of the protagonists of McCauley’s /Into Everything/:

“Like every freebooter, he had heard all kinds of cautionary tales about wild algorithms infecting the unwary and driving them crazy, and now something very like those storied horrors was curled up like a tapeworm in his brain. He wanted to believe Aunty Jael’s reassuring diagnosis, [but] could not help wondering how he would know if he began to think thoughts that were not his own…”

Now, perhaps, unlike Tony, you know some of the signs. You’re welcome. Always happy to help someone think about tapeworms in their brain.

On the one hand, I’m sorry about the mishmosh of comments on efficiency, order, power, simplicity, correctness, universality, and distance. On the other hand, I think that’s the *reality* of what drives people in the direction of Le Corbusier and Lenin: a mishmosh. Trying to separate out the desire for efficiency from the desire for order would itself be an example of Seeing Like a State, of trying to make a complicated reality “legible”. The dissatisfaction I feel with this episode is, I flatter myself, not because I didn’t get the script right but because my High Modernist eidolon is being stimulated by a reality it feels *compelled to fix*.

Next time, I’ll talk about ways of shutting up your eidolon, based especially on what real people do, both in the absence of the State and in its presence. I hope to help you design “schemes to improve your users’ conditions” without High Modernist failure.

I’ve decided to put off a discussion of how to fend off your corporate hierarchy when it wants to see and control and improve *you* against your will, at least until I cover Hirschmann’s /Exit, Voice, and Loyalty/.

Thank you for listening.