Boris Smus

interaction engineering

Seeing Like a State by James Scott

The most succinct version summary of Seeing Like a State this is what has been called The Authoritarian High-Modernist Recipe for Failure:

  1. Look at a complex and confusing reality
  2. Fail to understand all the subtleties of how the complex reality works
  3. Attribute that failure to the irrationality of what you are looking at, rather than your own limitations
  4. Come up with an idealized blank-slate vision of what that reality ought to look like
  5. Argue that the relative simplicity and platonic orderliness of the vision represents rationality
  6. Use authoritarian power to impose that vision, by demolishing the old reality if necessary
  7. Watch your rational Utopia fail horribly

On the other hand, humanity has clearly benefitted from modern institutions that scale. Scale requires optimization. Optimization requires simplification. But overly simplified schemes dehumanify people, converting humans into drones lacking agency and autonomy. My main question going into this book: how to strike the right balance?

Anyway, back to the book: Scott goes through many different examples of the above scheme in practice.

Scientific Forestry

Pioneers of modern forestry viewed forests as factories for growing trees:

"One should manage the forest so as to guarantee the highest sustained financial revenue, when all usable products are calculated in money." – Friedrich Wilhelm Pfeil, the first director of the Eberswald Academy of Forestry (1856)

This led to the unintended consequence of forest dieback. A century later, an observer wrote:

Many of the pure stands [of spruce] grew excellently in the first generation but already showed an amazing retrogression [decline in yields] in the second generation.” – Richard Plochmann, District Chief of the Bavarian Forest Service (1968)

City Planning

Modern planners believed that city planning can be done abstractly, without any regard for the specifics of the city's residents. Le Corbusier's vantage point is entirely external to the city, completely devoid of cultural context of the people he was building for. The resulting city was perfect for an average human that didn't exist. His plans were best seen from an airplane, and his sketches entering the port on or ship. Le Corbusier's concerns are visual.

"Houses are machines for living in." – Le Corbusier (1923)

This led to planned cities like Brasilia and Chandigarh. These cities were visually appealing from abstract vantage points, but not functionally livable. They worked for people only because of an inevitable departure from the architect's master plan, molding it to their actual needs.

One anecdote was particularly striking to me. Le Corbusier went so far as to design plans for Moscow that were judged too radical by the Soviets. Unfazed, he took the same plans and recycled them to use for central Paris! In contrast, Jane Jacobs in her book The Death and Life of Great American Cities, championed function over form, and walked in the shoes of a resident of the city.

Russian Revolution

Lenin believed that communism must be brought in through violent revolution.

"The organisation of the vanguard of the oppressed as the ruling class for the purpose of crushing the oppressors." - Lenin (1917)

He viewed the Vanguard Party as a machine for bringing on communism. Needless to say, the revolution led to a huge loss of life, and then even more in the repressions that followed. Conflicts with the Kulaks over collectivisation led to a terrible decline in quality of life for farmers and workers that Lenin was trying to help.

Rosa Luxemburg, a Marxist theorist and anti-war activist, and her spiritual successor Alexandra Kollontai, a Russian Menshevik, had a very different take on communism. Theirs was a nature-inspired view of social uprisings, strife, and revolution. This bottom-up view was diametrically opposed to Lenin's top-down view.

There are striking parallels with Jane Jacobs' critique of Le Corbusier. Is it a coincidence that in both cases, it is a woman's humanistic, nuanced, and intuitive tendencies criticizing the iron fisted absolution of a man?


Modern farming views the farm as a factory for producing food. The bigger and more specialized, the better. This trend was present both in the west and in the east in the early twentieth century, but it is illuminating to see the differences in approach. This illustrates a point Scott hedges on repeatedly, that high modernism by itself is not the problem, but requires authoritarianism to lead to disastrous results.

In America, large industrial farms led to bad outcomes, but these did not lead to massive famines or millions of deaths. In the east, the sheer scale of the operation led to disaster. Early Kolhoz were created by dividing whole swaths of Russia into a grid of 2500 hectare plots, and requiring quotas from each "farm" without much regard for whether the land was arable or not. Collectivization in the Soviet Union and China led to large scale famines and millions of deaths.

On the topic of agriculture, Scott explicitly mentioned one feedback loop I found interesting. Crops were modified to make them more amenable to mechanized picking. This led to a monoculture and a more fragile crop, and thus to increased pesticide use.

Applications to corporations

One of the major ideas in the book is legibility, the process by which modern states, over a long period of time, attempt to understand society, simplify it, and make it more managable. In Scott's own words, to

arrange the population in ways that simplified the classic state functions of taxation, conscription, and prevention of rebellion. Having begun to think in these terms, I began to see legibility as a central problem in statecraft. The pre-modern state was, in many crucial respects, particularly blind; it knew precious little about its subjects, their wealth, their landholdings and yields, their location, their very identity. It lacked anything like a detailed “map” of its terrain and its people.

It struck me that a lot of Scott's critiques apply to corporations just as much as they do to governments. In many cases, tech companies are in the business of taking an existing practice and digitizing it. Companies like Convoy and Cloud Kitchens, in an attempt to digitize the physical, are implicitly increasing legibility and imposing the same sorts of potential pitfalls that will be described in this book.

In particular, this involves the following:

  1. Understand the physical world
  2. Predict the future state of the physical world
  3. Control the future state of the physical world

This is exactly the same sequence of steps that most of the high modernist examples in this book undertake.

Recipes for success

Scott has a few suggestions which I succinctly summarize "be humble". He explicitly writes:

  • Take small steps, observe results, repeat.
  • Make sure the steps are reversible. The first rule of intelligent tinkering is to save all the parts. (Aldo Leopold)
  • Plan for surprises.
  • Plan on human inventiveness.

Scott uses "metis", a Greek word for know-how, or cunning, or chutzpah, to describe a lot of what's missing in high modernist ideology. Metis refers to the sorts of things that cannot be taught, but must be practiced, like riding a bike, playing an instrument. Scott argues this understanding of the practical is critical for any high modernist scheme, as the scheme is inevitably overly simplifying, so gaps emerge that need to be filled.

Metis has led to folk discoveries that were done by non-scientific method were often rediscovered by science in some form and formalized. Scott makes a sort of prediction that this traditionally derived wisdom is still far ahead of scientific discovery in some domains, such as poly-cropping for example.

One common theme across all of the high modernist failures is mechamorphism: everything is a machine. For Le Corbusier, the house was a machine for living in. For Lenin, the vanguard party was a machine for bringing forth communism. For practitioners of Forstwissenschaft, forests were factories for making trees, and farms are factories for making food. It's easy to make fun of, but maybe today's metaphors will be similarly flawed?

Seeing Like a State and AI

Speaking of today's metaphors...

In the spirit of Chesterton's Fence, it's worth taking a closer look at simplifying schemes that we take for granted. They may actually have adverse side effects. Can we undo them through a better understanding of the things that came before them?

High modernism often fails because it is overly reductive. I keep wondering: can we better understand complex, natural domains with modern technology? We've made really great progress in some areas like speech recognition. Can we apply AI to better understand complex domains without being overly reductive and prescriptive? For example, what if we built a gardening AI to take meticulous care of an organic polyculture farm?

Imagine a future in which a series of complex, domain-specialized AIs have been created in such domains as gardening and driving. We may well end up with systems that work super well, but are far less legible. What would the ensuing world look like? I'm imagining self driving cars with emergent swarming behaviors that look completely terrifying from the sky, but would be very functional and completely safe (as long as no humans were involved in driving decisions). I feel that we are very far from such a world, but it's fun to think about.