States See Abstractions

Dec. 10, 2019, 2:20 a.m.

The paradox of the abstraction is that the more general the abstraction, the less it can tell you, and so the more useless it becomes. The essence of abstraction is to unify that which is common; the more that can be unified, the less you will find in common. When you’re limited to working only at the level of the abstraction, this can be extremely limiting.

In Seeing like a State, James C. Scott puts forward the idea that in their pursuit of control, States have negative effects on the quality of life of the inhabitants. The basic premise that the State can only make decisions on what it can see (understand), the state wants to see more, that there is a whole world of information that the State cannot see (Scott calls this metis), and that because the State cannot see the metis, the State inevitably ends up destroying said metis, at great cost to its inhabitants. I won’t dive into a full breakdown, for a fuller understanding I highly recommend this summary.

Put another way; any large enough state has to abstract over its citizens/responsibilities, even if only because of scale. State’s are composed of processes, there's no room for a different process per citizen; they all must be treated equally, with the same process, no matter the irregularities. Scott refers to state to be any sort of government; going forward I’ll use the term organization, since everything that follows (I believe) applies equally to any sort of governing body, be it a corporation, neighbourhood council, or a literal government. I’ll be, if you’ll indulge me, abstracting over all of them.

Somewhere, most organizations have a person abstraction, be it citizen, employee, or resident. The pursuit of any abstraction is to be clean and pure, and to that regard minimal. A good abstraction contains only that which is necessary to make the abstraction sufficient, nothing more. Minimalism tends to be a good thing to aim for when designing such things.

The problem is that people are quirky little messes that explode the size of any abstraction attempting to meet the “good abstraction” criteria even something as simple as someone's name turns out to be pretty hard to do correctly.

Trying to abstract over something as complicated as basically everything, organizations will have faulty abstractions that inheritly fail to capture everything. Organizations only work at the level of the abstraction, because that’s the only level they can operate at. Thus, they necessarily fail to see what falls outside the purview of their shitty reduction.

H. James Harrington puts it well - “Measurement is the first step that leads to control and eventually to improvement. If you can’t measure something, you can’t understand it. If you can’t understand it, you can’t control it. If you can’t control it, you can’t improve it.”

In Seeing like a State, Scott goes to show that not only do organizations end up not improving the invisible metis, they actively seek to destroy it.

You can see this shift everywhere in consumer facing tech products. Consumers never have to interact with a person directly, as that would cost too much. No, customers have been abstracted over such that no human intervention is required; everything from onboarding to support has been automated. Google is perhaps the conical example of this, known jointly for its prowess in creating high performance software as well as it’s complete inability to deal with customers as people.

Once dependencies are properly abstracted over, it’s nice to make them fungible, so as to be less reliant on any specific implementation. If everything looks the same, why shouldn’t I be able to swap one for the other?

Scale begets abstraction, abstraction begets normalization, and normalization begets fungibility.

Fungibility begets competition, competition begets efficiency, and efficiency begets scale.

And throughout it all, what is lost? One large example I think is employee loyalty. Generalizing wildly, the days of working your whole career at one, maybe two companies, working your way up the corporate ladder are long gone. Corporations just don’t see an employees personal dedication in their annual reviews. I’ll grant you that it’s very difficult to measure a long-term employees total ‘value contribution’, but as of now all the little things that add up to make them an exceptional employee go largely unnoticed. People today change jobs much more frequently, as a direct consequence.

In fact, the fixed point of digressing employee loyalty has already been attained: the gig economy. Employees are so fungible that one can just sign up via an app, get work via an app, get paid via an app. The entire middle management layer has been removed, even if the employee wanted to move up the ladder, there is no ladder to move up.

Scott takes the whole process very pessimistically, that the State moves towards everything being the same cookie-cutter experience, everywhere, with a total ignorance to the human condition. The metis is lost forever, and with it, a little piece of humanity.

I like to take an alternate approach. In today's days of total surveillance and tracking, I like to see the good side: with better data, comes better decisions. Organizations may chase their bottom line relentlessly, but there’s no escaping the fact that customer satisfaction is the final line.

I guess only time will tell whether Orwell was right or not ¯\(ツ)