Ex Libris Kirkland

Buy it from Amazon

Subtitle How Certain Schemes to Improve the Human Condition Have Failed
First Written 1998
Genre Nonfiction
Origin US
Publisher Yale University Press
My Copy paperback
First Read September 06, 2024

Seeing Like a State



This book gets tossed around a lot in tech circles, and I think it does have a lot of interesting things to say about software. I often tell clients that in order to build software, we have to understand their existing processes - and usually that process is so fuzzy that it can’t really be replicated exactly. Thus to create new software, we will sometimes flatten or streamline or restrict what actions can be taken. Hopefully this is helpful, and worth the tradeoffs in flexibility.

This book is about how for a government - a state - to take action, it needs to be able to know things about its population, it needs to ‘see’ them. But the wild and wooly world isn’t naturally legible, and so states have to create artificial categories and boundaries and then force people into them. You can’t, say, tax a population if you don’t know how much they make. Or how many people there are! You can’t conscript an army if you don’t know where people live.

Thus, a top-down forcing of structure from above becomes inextricably tied to that structure being used as a tool of control. And when that top-down structure doesn’t fit or doesn’t make sense, your initiatives will fail. And Positive examples work the same way: you can’t, say, provide a food benefit to a population if you don’t know where they are. You can’t say ‘give a barrel of grain to every family’ if you people in your country have different sized barrels and wildly different notions of how many people a ‘family’ includes. Or: how do we make a vaccine available to anyone who needs it if we don’t have a realtime data source about where people are and how many medical professionals are available to deliver these?

Certain forms of knowledge and control require a narrowing of vision.” -> we can’t build a software system that accommodates ANY activity. Just the ones we plan for. We can expand ‘state’ to mean any organization that has some top-down control over its subjects. Corporations, clubs, etc. The whole university can’t ‘see’ your quality discussion or read your individual papers or see how much you’re learning and contributing to a community of knowledge. That’s ‘illegible’ to the Uni. The Uni CAN see your GPA and your attendance. Those are legible.

Beehives are a great example. Older beehives were kept in whatever structure the bees liked, and they make a big mass of comb. To get the honey out, you have to destroy the hive and hope your colony can regroup. BUT THEN we invented the modular modern hive system, with racks of frames that are just-far-enough away that bees dont try to fill the whole thing in with comb. Thus you can pull individual frames out for harvesting, the comb is all in order planes, and the colony doesn’t have to be moved. This is a top-down imposition of structure. We figured out ‘bee-space’ and built a system that works with it.

First big example in the book is forestry. A forest has many uses and many economically valuable things in it. But state forestry thinks first about board feet of lumber. If you strip out the forest and plant only the trees you want to harvest, spaced in even rows, it makes it easy to plan and forecast and track and account for that forest, and much more efficient. But you lose all the other value there: wildlife, acorns, mushrooms, brush that gets used for firewood, etc. You’ve lost a bunch of valuable product because that wasn’t legible to the state. Plus, after a generation or two even the trees you wanted will die because trees need a complicated ecosystem, not just soil+water+air.

“State agents have no interest—nor should they— in describing an entire social reality, any more than the scientific forester has an interest in describing the ecology of a forest in detail.”

He then compares this to land. A state needs to know who owns what land. But if you want to allocate land, you need a unit, and units are abstractions. An acre = surface area but doesn’t tell you if it’s rocky or boggy. What parts of that land are good for what? Irish measurement of ‘fair of x cows’ tells you about grazing capacity but not surface area. ‘Telling a farmer that he is leasing twenty acres is about as helpful as telling a scholar he is buying six kilograms of books.’

And there are names: you can’t just put out an arrest warrant for ‘Matt’. Maybe ‘Matt son of Doug’ or ‘Matt who farms on the church’s land’ But sooner or later your subjects are going to need permanent last names and you’ll end up with a last name that had previous meaning frozen in amber: “Matt Kirkland”. And then you’re going to graduate to giving people numbers, right? See Patio11’s https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/ of course.

The book goes into three big case studies of where a well-intentioned big plan from Above tried to force some order onto a wooly real-world society Below, and failed. The subtitle is ‘How Certain Schemes to Improve the Human Condition Have Failed’.
- forcing people into villages in Tanzania
- collectivizing farms in the USSR
- designing a big new city from scratch in Brasilia. Compare Le Corbusier . Or Robert Moses planning highways thru NYC neighborhoods, and Jane Jacobs street-level resistance.
Those histories are interesting enough but I think we have heard enough stories like this to imagine the details even without reading.

Scott then talks about how this artificial ordering from above then starts shaping reality on the ground. “A machine ins’t designed to harvest crops, the crop is designed to be able to be harvested by the machine.” Just think: google and search engines started so we could find information on the internet. And then everybody started trying to game google by shaping their content to fit the Great God Algo.

He draws out the differences between mētis and techne: mētis is the on-the-ground history and practice of doing something. Techne is the technical knowledge of how to do it. It sounds like these ideas should overlap, but they can be different. You can avoid the bubonic plague with your mētis: you know to avoid the dead, move out to the country, seal up the sick and don’t go near them. Your techne is because here is a bad miasma air. Your techne is wrong! 100% false! But your mētis knowledge is 100% effective, if you isolate yourself from the sick and get fresh air you won’t catch the plague. You can successfully know how to do something and also not know the real correct facts about how to do something.

Recipes are a classic example of the mismatch. ‘Whip the egg to stiff peaks’ (uh, what is a peak? How do you do that? How stiff?) ‘Heat the oil until it is almost smoking’ (assumes I have overheated enough oil to know when it’s going to burn).

But applying this to software design, some takeaways:
- world is complicated
- we have to simplify it to make software
- we should be careful and cognizant of what parts we flatten or exclude
- users who successfully accomplish things might still not understand why. And the reverse!

Noted on September 12, 2024

The best way to appreciate how heroic was this constriction of vision is to notice what fell outside its field of vision. Lurking behind the number indicating revenue yield were not so much forests as commercial wood, representing so many thousands of board feet of saleable timber and so many cords of firewood fetching a certain price. Missing, of course, were all those trees, bushes, and plants holding little or no potential for state revenue. Missing as well were all those parts of trees, even revenue-bearing trees, which might have been useful to the population but whose value could not be converted into fiscal receipts. Here I have in mind foliage and its uses as fodder and thatch; fruits, as food for people and domestic animals; twigs and branches, as bedding, fencing, hop poles, and kindling; bark and roots, for making medicines and for tanning; sap, for making resins; and so forth. Each species of tree-indeed, each part or growth stage of each species—had its unique properties and uses. A fragment of the entry under "elm" in a popular seventeenth-century encyclopedia on aboriculture conveys something of the vast range of practical uses to which the tree could be put.
… “
Elm is a timber of most singular use, especially whereby it may be continually dry, or wet, in extremes; therefore proper for water works, mills, the ladles and soles of the wheel, pumps, aqueducts, ship planks below the water line, ... also for wheelwrights, handles for the single handsaw, rails and gates. Elm is not so apt to rive [split] … and is used for chopping blocks, blocks for the hat maker, trunks and boxes to be covered with leather, coffins and dressers and shovelboard tables of great length; also for the carver and those curious workers of fruitage, foliage, shields, statues and most of the ornaments appertaining to the orders of architecture... And finally ... the use of the very leaves of this tree, especially the female, is not to be despised, ... for they wis prove of great relief to cattle in the winter and scorching summers when hay and fodder is dear... The green leaf of the elms contused heals a green wound or cut, and boiled with the bark, consolidates bone fractures.
…”
In state "fiscal forestry" however, the actual tree with its vast number of possible uses was replaced by an abstract tree representing a volume of lumber or firewood. If the princely conception of the forest was still utilitarian, it was surely a utilitarianism confined to the direct needs of the state.

Quoted on September 12, 2024

State agents have no interest—nor should they— in describing an entire social reality, any more than the scientific forester has an interest in describing the ecology of a forest in detail.

Quoted on September 12, 2024


Ex Libris Kirkland is a super-self-absorbed reading journal made by Matt Kirkland. Copyright © 2001 - .
Interested in talking about it?
Get in touch. You might also want to check out my other projects or say hello on twitter.