E17: James C. Scott’s /Seeing Like a State/, part one

Download MP3
An introduction to the core ideas of Scott's /Seeing Like a State/. Three examples. Nothing about software yet.

Welcome to oddly influenced, a podcast about how people have applied ideas from *outside* software *to* software. Episode 17: James C. Scott’s /Seeing Like a State/, part one.

The subtitle of Scott’s book is “How Certain Schemes to Improve the Human Condition Have Failed”. Most of the book is a series of case studies and his generalizations from them, plus speculation on how such failures can be avoided.

He focuses on nation-states as the actors responsible for the failures and argues that there are four necessary elements.

First is an administrative simplification of nature and society. For now, think of the distillation of a complex real-world situation into summary statistics or, let’s just say, a domain model. Such simplifications are necessary for a modern society to work well. But getting them wrong sets the stage for problems, or perhaps catastrophe.

Second is a “high modernist ideology”. That ideology is “best conceived as a strong, one might even say muscle-bound, version of self-confidence about scientific and technological progress, […], the mastery of nature (including human nature), and, above all, [confidence in] the rational design of social order [based on] the scientific understanding of natural laws. […] More about that later.

The third element is “an authoritarian state that is willing and able to use the full weight of its coercive power to bring these high-modernist designs into being.”

The fourth element is “a prostrate civil society that lacks the capacity to resist these plans.”

Scott is relevant to us in software for two reasons. First, the corporation you work for is essentially authoritarian in that, until you decide to leave it, you have to obey orders or at least pretend you are. And, historically, the civil society of employees is fairly prostrate in the face of that authority.

Beyond that, computer programmers are often themselves authoritarian. If someone’s boss tells them they have to use your app, the design decisions embedded in it just *are*, in part, a design of part of that user’s social order, a part of their life. It might be good to know what abusive high modernism is and acts like, so as to avoid acting that way yourself, and to know how people react to imposed social orders, so as not to fail because you believe users will behave as you think they should rather than as they will.

This first episode will explain the core of Scott’s ideas. I hope that the next applies them to software. A warning: I am finding this book hard to turn into podcast episodes. If this episode is rough, I apologize. If the next deviates from the plan, I again apologize.


Imagine you’re the king of Prussia. You own your kingdom. It includes a whole lot of forest land, and lumber is a significant chunk of your income. In order to look at the economics, let’s say a plot of forest land takes 50 years to recover after you chop the trees down for their lumber. So, to leave your descendants a sustainable source of income, you only chop down 2% of the forest every year.

The problem is that the amount of lumber you can get from a random 2% of a forest varies really widely – a forest in a river valley is different than one in a higher plain is different than one in a mountainous foothill, meaning your income fluctuates a lot from year to year. Fortunately for you, you’re a ruler toward the end of the Age of Enlightenment, which was all about applying rationality, science, and mathematics broadly, including to human affairs, including the management of kingdoms.

So you enlist people who develop standardized tree sizes, and train other people to recognize them. Then those other people go counting and classifying your trees. With the resulting data, plus tables of how much usable lumber can come from certain sizes and ages of trees (calculated using the geometry of cones), a forester can easily and accurately estimate the total yield of a forest or a section of forest. You, the monarch, now have the ability to adjust your harvesting to produce roughly the same amount of lumber each year, sustainably.

However, the forest is still painful to work with, harvesting-wise. The people doing the measuring have to tramp around through underbrush and over fallen logs to do their job. And that clutter makes the actual logging really awkward and expensive.

So a sensible thing to do is to replant a harvested area with rows of trees, nicely spaced and with corridors between the rows, to make harvesting easier. And send in people to keep the underbrush clear and drag out fallen trees. Not only does that make managing the growing forest easier, it looks a lot *tidier*. You, as a product of the Enlightenment, are super fond of tidiness.

And while you’re at it, why replant just any old trees? Norway Spruce is more hardy than your average species of tree, grows fast, and produces good wood. So let’s plant only Norway Spruce.

We nowadays know that “monoculture” is a scary word, so we can predict what the scientific foresters of the 1800s didn’t. Pests specific to Norway Spruce had a grand old time. The monoculture’s ecosystem was rather barren – not offering much purchase for the lifestyles of animals that might have eaten those pests. Woodpeckers, for example, eat bugs, but they need hollow logs to live in, and your foresters hauled out any dead trees.

And separated trees all in a row can get knocked down like bowling pins if there’s a big enough storm.

*And*, it turns out all that underbrush that got dragged away is important for keeping the soil fertile. Because of that, the *first* harvest of a scientific forest was great. The next harvest was down 20-30%. And things looked to get worse after that.

So the scientific foresters had to introduce… kludges. Boxes for woodpeckers to colonize, replacing hollow logs. Ant nests were reintroduced, and children were used to tend to them. Spiders were imported and let free. And so on and on.

I like to think of that as trying to introduce *just enough* variety to keep the forest going. I’m guessing modern foresters have gotten even better at balancing control with diversity, but I don’t know and Scott doesn’t say.

In addition to trees, you as King have another crop: people. People pay taxes, and people can be conscripted when you need soldiers. From the king’s point of view, the problem with people is you don’t know how many of what type you have – it’s really the same problem as with trees.

Consider the issue of what we’d today call property taxes. Let me set the scene.

In a particular village, certain families might have the right to plant certain plots of land, but only with certain crops. Every seven years, though, the rights are redistributed to account for changes in the sizes of families and the number of able-bodied adults in each family. After the main crop is harvested, the land is opened up for grazing, but the number of animals one family can graze depends on the size of the family. Everyone can gather wood for their own use – unless, that is, they live next to a scientific forest, which doesn’t have any downed trees for them to gather – but no one can sell wood. The fruit from a tree planted by a family belongs to that family, no matter how long ago they planted it or where the tree even is. But harvest quickly! Anyone can gather fruit that’s fallen to the ground. And if a tree falls or is chopped down, the trunk belongs to the family that planted it, the branches belong to their immediate neighbors, and the twigs and leaves can be taken by any poor villager who wants them.

Now, pretend you’re one of the king’s officials. Your job is to figure out how much property tax each family in this specific village should pay. Oh, and keep in mind that once you finish with this village, you’ll have to do the same for the next village – which has *different* customs for all these things.

In such a situation, it’s impossible for a central authority to know how much to tax, so the job had to be delegated to someone local, a local elite who was responsible for collecting tax and forwarding it to the king. But those elites had every incentive to cheat, to undercollect from their neighbors (and perhaps skim something off for themselves).

Alternately, you could let people – so-called tax farmers – bid for the right to collect taxes. The deal is they promise to deliver a certain amount of tax every year. If they can squeeze more out of the peasantry – and they very much expect to – they get to keep the excess.

In addition to the problem that tax farming makes your government unpopular and can lead to peasant revolts, you as king are at a serious informational disadvantage. You know way less than the tax farmer about what a sustainable tax is, and the tax farmer has every incentive to bid low and keep a larger excess.

If your state is to increase its tax revenue, it’s going to have to simplify these fluid and local property rights into something more measurable and standardizable: modern property rights. Each family will be assigned a specific plot of land. They pay tax on that acreage. Oh, the taxes might vary according to presumed average yield of the single crop that’s been decided should be grown there. And the yield expectation might be adjusted according to things like soil quality or availability of irrigation. But such measurements inevitably ignore far more than they include. As Scott says, “The fact that a field designated as growing wheat or hay might also be a significant source of bedding straw, gleanings, rabbits, birds, frogs, and mushrooms was not so much unknown as ignored lest it needlessly complicate a straightforward administrative formula.”

Here’s a core distinction. Someone who wants to make a *living* off the land, wants rough, idiosyncratic, and context-sensitive measurements. They’d much rather know that the yield is consistently between 4 and 7 baskets than that the average yield is 5.6 baskets. The worst case is much more important than the average when it comes to worrying about whether you might starve some year. And farmers don’t care about acreage – they want to know how many days it will take to plow or weed a plot, taking into account the local practices for plowing and weeding.

Generally, the administrators who forced private ownership on people used to collective ownership didn’t care about such things. Remember, the purpose here was not to accurately understand what the property is worth to the people who live there, but to understand – and maximize – what wealth can be *extracted* from those people. In many cases of “rationalizing” agriculture, throughout history, the effect was to reduce total value but increase the value extracted by the state.

In Scott’s book, this tension between what helps the people on the ground, and what helps the sovereign, is always present. He assumes that people don’t typically want to have their lives “improved” and will resist.

So let’s finish this first episode by looking at three of Scott’s four central terms.

The first is “synoptic”, which is related to the word “synopsis” or a summary. But Scott’s use picks up on some connotations of the dictionary definition. The first is that a synoptic view is a *comprehensive* view, a summary of all – and only – the important parts.

Another connotation of “synoptic” is “giving an account of events from the same point of view”. That point of view is that of the sovereign. The point of view of the people on the ground is not relevant. What the sovereign cares about – like the amount of lumber – is what counts, not the points of view of people who lived near the forests.

Another of Scott’s key words is “legible”. His metaphor is that the messiness of reality is *condensed* and *simplified* to something like a map. Maps don’t completely describe the landscape; they are purpose-specific. A road atlas is focused on showing how you can get from one place to a far-away place. A topographical map shows elevation because that matters if you’re trying to hike from one place to another. People driving in cars don’t care about elevation, mostly, so road atlases mostly don’t show it.

Another way to think about Scott’s use of “legible” is that it adds a layer of indirection. Administrators don’t deal with the facts on the ground; they deal with a formal representation – a narrow representation – of those facts. It’s as if you’re reading a description of a scene in a novel; you know what you’re reading leaves a lot out, only including things that support the author’s purpose going forward.

Both “synoptic” and “legible” have something to do with seeing, and there’s a strong metaphorical component to them. The administrative state generally thinks of itself as *above* the society, looking down on it – a sort of God’s-eye view. Or it envisions itself as looking outward from the center.

This has consequences, as the aesthetics of orderliness or elegance can dominate usefulness. A famous example is the city of Brasilia, built from scratch in 41 months, starting in 1956. From above, it looks something like a bird or an airplane. Its plan is really much prettier than old cities that just grew every which way. However, it was – and maybe still is – famously a bad place to live. For example, “There *is* a square. But what a square! The vast, monumental Plaza of the Three Powers […] is of such a scale as to dwarf even a military parade. In comparison, Tiannanmen Square and Red Square are positively cozy and intimate. If one were to arrange to meet a friend there, it would be rather like trying to meet someone in the middle of the Gobi Desert. […] And if one did meet up with one’s friend, there would be nothing to do. […] This plaza is the symbol of the state; the only work that goes on around it is the work of the ministries.”. But it sure does look cool from an airplane.

While we’re picking on Brasilia:

“Brasilia has few landmarks. Each commercial quarter or superquadra [that is, group of apartment buildings] looks roughly like any other. The sectors of the city are designated by an elaborate set of acronyms and abbreviations that are nearly impossible to master, except from the global logic of the center. Holston notes the irony between macro-order and micro-confusion: ‘Thus, while the topologies of total order produce an unusual, abstract awareness of the plan, practical knowledge of the city actually decreases with the imposition of systematic rationality.’”

This weird reliance on aesthetics will come up again in the next episode.

The final term (for now) is “High Modernism”. This is necessarily somewhat vague because we’re talking about periods of history, and those don’t have clear boundaries. Roughly, plain old modernism is the product of the Enlightenment and the industrial age. It’s associated with the idea of both technological and political progress, that the future will be better than the past, and that human action can *make* it better. It also sees a trend toward increased individualism and the idea of self-improvement. As political institutions like the nation state take hold, the idea that people can improve themselves tends to shade into the idea that the state should improve them. For example, in the Russian Stolypin reforms of 1906-1914, quote, “The state officials and agrarian reformers reasoned that, once given a consolidated, private plot, the peasant would suddenly want to get rich and would organize his household into an efficient workforce and take up scientific agriculture.” That didn’t work, as the Czar and his ministers found out in WWI. Later, the USSR would double down on creating what they called the New Soviet Man, with even worse results.

Modernism shades into *high* modernism. High modernism has a strong fetish for technology and science; it tends to assume that an elite knows enough, right now, to construct a better, even utopian, future. (It doesn’t care much about the present and tends to think the past provides only negative examples.) The elite assumes that they know better than the common people how those common people should live their lives, and what the social order they live within should look like.

I find it hard to express the difference between the high modernist and modernist attitudes toward people. Here’s the best way I’ve found to put it:

- when Prussian officials rationalized the forests, they really hurt the peasants. Peasants no longer had underbrush for their animals to graze on, they had nothing to hunt, they couldn’t gather medicinal plants, and so on. If an official had been informed of that, his attitude would likely have been, “Huh. Sucks to be them.” The state has the model of reality that matters to it, and it’s indifferent to what falls outside the model.

- the high-modernist ideologue, informed of the same thing, would have the reaction, “They’re just *wrong* to care about such things.” The ideologue believes he has an abstract, global, platonic, scientific, and *complete* list of what should matter to the people whose lives he’s rearranging. Depending on the situation, he may ignore their desires or adjust his plan to “fix” the people. In the planning of Brasilia, the desires of the future residents were ignored. The Soviet Union put a lot of effort into fixing the peasantry.

It’s important to note that we’re all beneficiaries of modernism and even the less arrogant versions of high modernism. Public health is a case where a scientific elite used science to improve people’s lives, in part by, um, fixing people. The US Centers for Disease Control is in Atlanta, Georgia, because malaria used to be endemic in the US south. Now it’s not – a case where “seeing like a state” worked well.

And, while scientific agriculture was a disaster in the Soviet Union, many of the same ideas - like a concentration on crops that reward scale: maize and soybeans - have worked well in the US Midwest.

Scott is primarily interested in cases where “seeing like a state” didn’t work, and so am I.

Here are two relevant quotes that hint at why high modernist “schemes to improve the human condition” often fail.

The first is from Herbert Simon, quote

“Administrative man […] is content with the gross simplification because he believes that the real world is mostly empty – that most of the facts of the real world have no great relevance to any particular situation he is facing and that most significant chains of causes and consequences are short and simple.

The second quote is from Scott:

"If I were asked to condense the reasons behind these failures into a single sentence, I would say that the progenitors of such plans regarded themselves as far smarter and farseeing than they really were and, at the same time, regarded their subjects as far more stupid and incompetent than *they* really were."

Fortunately, tech people would never think such things, so we’re safe from failure.

Thank you for listening. Just as a note, I’ve set up a Mastodon instance for “Oddly-influenced”-type discussions. It’s social.oddly-influenced.dev. I have some craving for a community of fellow widely-read dilettantes, and maybe that instance could be it. (You can find a clickable link in the show notes.)

See you next time.

E17: James C. Scott’s /Seeing Like a State/, part one
Broadcast by