Saturday, June 29, 2024

seeing like a state machine

I've been into automation games lately. This is a style of game where instead of punching trees to get wood, you get robots to punch the trees for you. Typically the output of one automation goes into the input of another to make increasingly complex supply chains. The goal of the game is to balance your inputs and outputs for maximum efficiency while moving towards more sophisticated products. The key to these games tends to be monitoring your inputs and outputs, which makes interface elements like production charts incredibly important. When I picked up Nova Lands, a smaller-scale automation game, I was boggled when there were no production charts at all. 

Seeing Like A State is a book detailing failures of top-down administrative beueararacacraacy (Cards on the table - I will never learn how to spell this word). The book takes a few case studies of elements where The State, such as a king, wants to know how many ships they can create in order to judge their ability to go to war. In order to know how many ships can be built, the state must be able to tally the amount of lumber that can be produced. In order to know the amount of lumber that can be produced, they must know the square footage of their forests and how much lumber can be obtained per square foot of forest. However, forests are not homogenous. Forests can be dense or light, and their boundaries might be marked with scrub or they might have clearings. Wouldn't it be easier if we could simply chop down the messy, unmanaged forests and replace them with scientifically planted forests containing trees in neat rows, allowing the state precise measurements of square footage and yield per square foot, thus lumber yield, thus the number of ships that can be built, thus the wartime capacity of the state? 

Every management game puts you in the perspective of A State, in one way or another. In Civilization you literally embody the undying avatar of a civilization, in SimCity you are the omnipotent mayor, in Dwarf Fortress you are the unseen commander giving orders to make buildings and workshops. As The State, you must know the sum total of your capacity in order to make decisions - can I afford this new building? Can I overpower my foe in a war? Do I have enough stockpiles to survive the winter? Certainly it is possible to play all of these games poorly, to reject knowledge and therefore efficacy. Yet mastery pushes the player towards certain directions, most of all in the automation game. The automation game is most concerned with efficient inputs and outputs, making the most of your forests to supply your lumber yards to supply your shipwrights. Mostly this is achieved by skyrocketing costs later in the supply chain. For example, in Satisfactory, it's fairly easy to produce Copper Wire: 1 copper ore produces 1 copper ingot which produces 2 copper wire, 15 per minute. Efficiency does not matter much here. By the time you need to produce motors, you need 240 units of wire per minute. Inefficacy strains the limited number of copper mines, the highly constrained power supply, and the wait time for prerequisite production items to flow through the chain. Therefore, monitoring throughput and efficacy is an important part of playing Satisfactory. The game includes many, many elements to help you understand the throughput of your production line. 

Nova Lands has no charts or graphs to show throughput. The best you get is little stockpiles at your factories to show the surplus accumulation of inputs or outputs. You must jet your avatar across islands, eyeballing where things are piling up or where factories are idle. I rebelled furiously against this. This was a waste of my time! These facts could have been consumed neatly and presented in an organized fashion from a single interface! My avatar could be sitting with his feet up on his throne, perusing reports and dispatching orders to shore up productions! 

Scientific forestry failed in its first incarnations. Replacing the natural growth of forests, messy as they were, with neat lines of trees did not account for the web of life that relied on those interplays and helped maintain the health of the forest, as well as provide food for the people living on the outskirts. The lack of diverse stock meant a single disease could fell an entire grove. Seeing Like A State, through this example and others, shows how a top-down view often ignores these facts-on-the-ground. One solution the book arrives at is "local control". Who knows the relationship between the trees and the rest of the forest better than the beurarararacararat? The forester who works it. The person who is present, lives the relationships that already exist, and suffers the consequences of homogenization first. 

Real life is entangled, but digital life only has the functions we give it. Nova Land has no complex web of life sustaining it, or consequences for translating the entire land into bleak industrial output. It would be a mistake to simply conclude that charts are a tool of the state, and by removing them and forcing me to hop from island to island, Nova Land embodies the concept of "local control". The inputs and outputs here are still neat homogenous bricks, easy to count, every one ultimately interchangeable for the other. However, Nova Life is designed - it controls the menu, therefore it controls the choices. It does not have charts or graphs and as a result, my time was spent personally monitoring the inputs and outputs of each factory and island. By thinking about my avatar as investing time in the islands, I was able to reframe my interaction with Nova Land's interface from "This is missing something I am used to" towards "I am getting a new perspective by having this element taken away from me". I stopped thinking about what I wished it did differently, and started engaging with what it was doing for me.