OSD 322: Building the government’s eye of Sauron
On Palantir, and government competence tech vs. government empowerment tech.
There are many schools of thought about how a government should use national economic statistics. Are trade deficits a bad thing? At what rate should GDP grow? How can you make it grow faster? People build careers answering these questions.
There is one approach to national economic statistics that you rarely see in governments: do nothing. In fact, don’t even collect the data. Having it would tempt you to do something.
John Cowperthwaite was the financial secretary of Hong Kong from 1961 to 1971, and that was basically his policy. A mentee once asked him to name his proudest act in office, and Cowperthwaite replied, “I abolished the collection of statistics.” Now, that’s not exactly true, because Cowperthwaite did keep some basic economic data. But overall his tenure was characterized by the British government begging him for the kind of data it was used to in governing its colonial possessions, and him refusing to collect it. All the while, Hong Kong’s economy boomed.
Ok, put that on the shelf for a bit.
There was a good ol’ internet dustup last week when Paul Graham tweeted this, referring to the US government using Palantir’s software to find the physical location of people it wants to deport.
That naturally got a response from people who believe in Palantir’s mission, most notably from this longtime employee:
This is a genuinely complicated issue. We focus on civilian defense, but this stuff is relevant there too, because we’re talking about generalizable tools. Anything that civilians can buy, the government can buy too. If it can be weaponized, they will weaponize it. Stopping the government from misusing a general purpose tool is no easier than stopping a doomed teenager from putting an auto switch on their Glock.
If you’re reading this in search of a magic solution, you won’t find it here. There are two statements about this whole thing which seem both indisputably true but also in extreme tension with each other. That tension is where the complexity of the issue comes from:
Palantir’s software helps the government do its legitimate functions much better. This is a very good thing. When you interact with the government, you surely want them to be competent.
If the government ever puts us in camps, they’ll use Palantir’s software to do it.
Does that mean Palantir is bad? Maybe, maybe not. Think of it another way: the guards in the camps will surely have Glocks on their hips. Does that mean Glock is bad? Maybe, maybe not.
People might recoil at #2 above. Maybe the idea of camps in the US is hyperbolic (excepting the time it happened). Will tyranny ever come here? One way to test that is to see what kinds of tyranny people already tolerate without issue:
Were COVID lockdowns tyranny? Is it tyranny that 20-50% of your earnings are confiscated every year? Is it tyranny that the Supreme Court allows Congress to reach thousands of miles away to regulate what you do on your own property? Is it tyranny that the federal government operates with impunity despite most federal law being unconstitutional by any reasonable reading of the Commerce Clause? Is the status quo of immigration law tyranny? If you’ve ever developed land or gotten a building permit, is the municipal permitting raj tyranny?
The point isn’t to litigate whether each particular example is or is not tyrannical. That’s up to each person to decide. The point is just that when you add it all up, the US is already today awash with government action that (a) a large fraction of the country genuinely finds tyrannical and (b) is conducted with tools that private companies eagerly provide to the government.
The kneejerk answer is “Companies should simply refuse to work with the government on anything tyrannical.” And sure, they should. But tyranny is usually only recognized as such after the fact, and even if people do recognize it in the moment, disengagement is harder than it seems.
You don’t even have to go all the way to exotic data analytics software or cutting-edge military hardware to illustrate this point. Let’s go much simpler. The analysis can begin and end with Microsoft Office.
If the government is doing something today that you find tyrannical, you can bet they are using Word and Excel to do it. Is it really tenable to say that anyone worried about those tyrannies shouldn’t work for Microsoft? If that’s the standard, there’s hardly anywhere in the country you could work. Forget any tech company or consulting firm. Forget law. Ditto any big retailer, any automaker, or any shipping company. Do you work for a drywall manufacturer? Your sheetrock is probably in some government buildings, so you’ll have to resign too.
Firing the government as a customer might work in extreme cases, but those cases will tend to occur during exactly the sort of emergencies when the mob is happy to arm the government with whatever tools it wants. So one company will stop working with the tyrannical government, five others will rush in to win the contract, and we’re right back to where we started. We need a better framework.
The wisdom of Cowperthwaite’s approach was that it didn’t rely on people making the right decision. Instead he tried to make it structurally impossible to make the wrong decision. Antonin Scalia captured a similar sentiment in some Senate committee testimony we like to quote, where he said that the reason Americans retain some freedom isn’t the promises in the Constitution — it’s that the framers made it structurally very hard for the government to do anything, because government doing things is what destroys freedom.
So it’s likely the solution to our problem isn’t to hope people and companies make the right decisions. That’s gravy if it happens, but as Milton Friedman said, a good system makes it so even the wrong people will do the right thing. You need a structure that boxes people into the right decisions.
What’s a structural solution to the question of when and how to equip the government? Here are a few ideas:
Distinguish between government competence tech and government empowerment tech. When you interact with the government, you definitely want them to be competent. That means using great software, having good equipment, and doing all the things you’d expect from an excellent private company. Tech that enables that is a good thing. But be wary of tech that focuses on empowering the government relative to individuals.
The government can have whatever it wants, as long as I can have it too. There is no viable near-term path to applying this to heavy military equipment, but why start with far-fetched examples? Let’s start with basics. And not just small arms, which people naturally think of first. Deregulated drones and comms are even more important.
Beware proprietary government data sets. There are some data sets relevant for the government’s internal operations that will inevitably stay private. Lists of employees, a database of the Army’s tanks, and so on. But any time the government is compiling data en masse about individuals, or claiming root access to a private company’s processes or data, that is a One Ring that someone is eventually going to try on for size.
The most important thing is what will lead to structural solutions in the first place: a culture that prizes freedom. These discussions are complicated, and there isn’t a ready answer. But if we keep having them and keep the culture oriented towards freedom, we’ll be moving in the right direction.
This week’s links
If the cops ND your gun while arresting you, apparently you can be charged with aggravated assault
From a weird case of a man arrested in a train station.
About Open Source Defense
OSD Capital
We invest in civilian defense and the tech that accelerates it. Reach out if you’re working on things in that area.
OSD podcast
In-depth interviews with outstanding founders and builders in the civilian defense industry.
The company store
Grab a t-shirt or a sticker.
Discord server
The OSD team is there along with lots of subscribers. Become a paid Substack subscriber to join the chat.
Simply as a exercise in posterity-- according to the extrapolation of the paragraph : where would one find it a best fit to earn money or work in? Would really help me out with that answer! Big love!