I heard it was you
Talkin’ ’bout a world
Where all is free
It just couldn’t be
And only a fool would say that – Steely Dan

A hero is someone who understands the responsibility that comes with his freedom. – Bob Dylan

Here’s a ramble that’s been hanging in my drafts for a while: Why do so many tech folks identify as libertarian? What do they mean? Is politics ripe for tech disruption? It’s a brand new year, and all old posts must go!

Now, I tend to lean, as the cliché goes, socially liberal, fiscally conservative, market-oriented. So I have some affinity for libertarian thinking. I tend to like that libertarians emphasize human rights, think outside the box and reason from first principles. But a lot of the people who identify as libertarian just take it too far. What gets me going is the person who, whenever something is messed up, thinks “OMG, if the government would step out of the way, a market solution would appear and work so much better!” … Even for activities which historically have a strong government role, roads, schools, money and banking, safety standards, pollution, etc.

I’ve written (here and here) about why I’m not a fan of that strand of libertarianism.

If you think the notion of immaculate, spontaneous order is a straw man, oh boy, just go to a Gary Johnson event and hear him questioned about whether it’s OK for the state to issue drivers licenses and speeding tickets. Not to mention debates about whether the Civil Rights Act of 1964 is a worse infringement on liberty than Jim Crow and segregation, or it’s moral for the government to tax people to save them from an asteroid impact, or whether parents should have the legal right to not feed kids and let them die. These aren’t just thought experiments to them, the libertarian scene is a hotbed of non-seriousness to downright kookiness.

I like freedom as much or more than most. But the idea that freedom is, in and of itself, a sufficient condition for desirable outcomes for everything has never seemed a compelling, complete theory that can deal with public goods, coordination problems, financial stability, most of the things we consider economic policy. It just begs the question, it’s the ‘assume a can opener’ of political ideologies. Once you give everyone freedom, you still have to design a market system and system of laws to solve social problems, define and assign property rights, decide what gets traded and how1.

Policies are market designs. Markets are sophisticated mechanisms, complex social institutions that have highly co-evolved with economic and political frameworks. There is no such thing as a pure free market, just as there’s no such thing as a pure human ‘state of nature’. Free markets depend on laws, cultural norms, technology which are like David Foster Wallace’s water. They may seem natural and invisible because we’ve internalized them…but try explaining them to an alien or an uncontacted Amazon tribe. Laissez-faire is not even wrong, it’s not even a thing. Free markets are games within a game, they are institutions humans create to serve objectives, they have to obey natural laws but they don’t exist in a social and cultural vacuum, they are intensely developed to be fit for purpose.

And market designs involve values and choices. Freedom is not the only human value that matters. Policy involves difficult tradeoffs between freedom, some notion of effectiveness, and some notion of fairness. Here’s a fancy graph!

Venn diagram of ideologies
(3D version which seemed more logical but turned out pretty hard to read)

That’s my political science ‘theory of everything’. (Inspired by Keynes) People organize to pursue their individual and group values2. They can choose to maximize different values. One universal value is freedom, or maximum power to the individual. Another universal value is fairness within the in-group according to some value system… each accorded status, power, material reward in accordance to their contribution and merit, with no one taking undue advantage of anyone else. Another value is ‘efficiency’, which is a catch-all for other possibly non-universal group values: economic efficiency and growth, national power and world supremacy, religious or sectarian values.

Extreme pursuit of individual freedom leads to libertarianism. Extreme pursuit of fairness leads to communism. Extreme pursuit of narrow group values at the expense of freedom and fairness leads to fascism.

Are tech types libertarian in that extreme sense of primarily maximizing freedom? Clearly not!

The vast majority of Silicon Valley political contributions go to Democrats. A tiny sliver of CEOs identifies as Republican, sometimes keeps it secret, or wears it as a badge of contrarianism. Someone like Travis Kalanick used Ayn Rand as an avatar, but loves Obamacare, which lets gig-ers get healthcare without putting the onus on their employer/platform.

The simplistic stereotype is that the tech elite are Randian technocrats who just want the government to get out of the way while they solve the big problems that confound the politicians, lobbyists, and bureaucrats, and make the world a better place.

But I think the proper analogy is that Silicon Valley techies view government as code, the operating system that defines the operation of society…an operating system maybe overdue for disruption and a major upgrade. And Silicon Valley types think they know code, know complex systems, and are just the ones to do it.

Let’s explore the analogy a little.

• Government defines rules between the people, corporations, the administrations, foreign policy. An operating system defines rules, protocols, APIs between the users, administrators, programs and processes, hardware, networking with other computers.

• Government defines a market design that allocates public goods between competing interests. An operating system defines rules that allocate hardware (e.g. CPU time, peripherals) between different processes and users.

• Like an operating system, you want government to have as small a footprint as possible, use as few resources as possible, and give users and developers maximum freedom to fully utilize all the computing resources at their disposal.

• Like computers, economies have grown bigger, more powerful, and more complex. Both operating systems and governments have taken on greater roles over time. Add multiprocessing, now you need a good scheduler, memory protection. Add multi-user and networking, need much better permissioning and security.

As systems get better and more complex, users expect more and more from the OS in terms of services, coordinating and scheduling complex tasks, security, supporting complex peripherals. So, as the system gets better, the percentage of work done by the OS vs. individual programs tends to grow, and the resources it uses grow as well.

That’s an essential paradox of operating systems and government. A better society is one where individuals are maximally empowered, and can also coordinate with others in complex ways. It’s a hallmark of improving civilization to give more people access to a more complex network of interactions, because they are safe, they have good norms of behavior and trust each other, they have clear laws well enforced, they have access to basic services. The more complexity is supported, the better the society. More complexity and interaction tends to mean more complex laws and regulation, and more shared resources supported by government.

That government is best which governs least, but as society grows more complex, the bar will tend to rise.

• In code, you want to abstract problems into modules that solve parts of a problem well. In politics you have the concept of subsidiarity: national, state, local governments; solve every problem at the lowest level of government that can effectively deal with it.3

• The qualities that make good code are agreed upon…efficient use of resources, easy to use, simple to understand, easy to read and maintain. But code involves hard tradeoffs…fast, good, or cheap? Pick any two. What actually is good code is partly subjective. People get into religious wars about methodologies and languages…agile vs. extreme…functional vs. object-oriented, etc.

Premature optimization is the root of all evil. Government changes very slowly, tech changes very rapidly. So any optimization in government, i.e. writing really detailed regs applicable to the current state of technology and the economy, is generally a premature optimization. Therefore government is the root of all evil? Well anywhere there’s something really wrong it’s probably a government failure, by definition. But it’s a catch-22, you don’t optimize, government is bad; you optimize, and eventually you end up in an unplanned-for situation where the optimization makes things worse.

Checks and balances slow things down. They’re sort of like bias and variance. You want to tune your system so that it adapts to change…but if it responds too quickly you can get instability and overfitting. You want your government to respond to the people and to changing circumstances, but not pivot abruptly on every whim. Direct democracy can be a double-edged sword. <cough>Brexit, Proposition 134<cough>.

• Sometimes you need to patch code pretty quickly and you accumulate technical debt, expedient solutions which need to be more carefully implemented later, or create maintenance and other problems in the long run. Hard cases make bad law…the solution that seems just in a complex case doesn’t always generalize. You don’t want to optimize code too much for the current use case, you need to be flexible. And you need laws that reflect universal principles. Sometimes worse is better…a simple, cheap solution everyone understands is better than a highly optimized, over-engineered one.

The euro is the ultimate in political ‘technical debt’. Build something that works today, even though it will need major re-engineering to be robust over the long term, and hope you can implement version 2 in time to avoid collapse.

• As you add complexity, a well-designed system can iteratively get better and better. Up to a point. There comes a tipping point where it becomes impossible to maintain and iterate on. Sometimes you accumulate too much technical debt, or change is just too fast, and you need to start over from scratch. Security exploits, bugs multiply. The same may be true of society, the tree of liberty, etc.

• Any computer system can get hacked. When you have a good market design, it’s a good basis for powerful interaction and coordination. But all systems can be gamed and exploited. Unless people are mostly decent and bad actors are sanctioned, you can get moral hazard and a tragedy of the commons.

• We don’t write systems or build hardware in which the components are told, here’s what we want to do, you guys work out the protocols5. APIs and protocols are specified in gory detail using things like RFCs.

• You can write awful systems in any language, and you can write pretty good systems in any language. It’s the same with ideologies. There are countries with moderately socialist orientations that work pretty well, and there are socialist countries which are disasters. There are countries with mostly market-oriented solutions that work pretty well, and there are free-market countries of weak-state and strong-state varieties which are disasters.

Personally, I’ll take Sweden over Pinochet’s Chile. In my opinion there are no countries at the extremes of any axis that work well over a long period of time. Somalia, with no working government, is in some sense the apotheosis of small government and maximum private freedom, and not exactly paradise on earth.

Ideologies are like development methodologies, people get pretty worked up about them, you need some methodology, but any reasonable methodology that isn’t applied so strictly it gets in the way is probably OK, and having decent developers is more important than which methodology they use.

If you’re concerned about Obamacare trampling on liberties, and not about people unable to get care, being bankrupted, facing impossible choices and preventable deaths, you might be an architecture astronaut caring more about buzzwords and ideological purity than actual solutions, outcomes, lives. The person who is free from government interference, because they’re dead, doesn’t notch it up as a win for liberty. The user wants code that works, and if it doesn’t work it doesn’t matter if it was made using scrum or Ruby.

Ultimately, I think the Silicon Valley brand of libertarianism is really the cult of disruption; healthy skepticism of government as premature or partial optimization and technical debt; nerd suspicion of wooly MBA/JD pointy-headed boss types; desire to empower people with tools, knowledge, ability to make their own choices, build their own solutions; when that requires a strong guiding hand in the form of code, or government intervention (education, net neutrality), so be it.

When you translate the rules of society into code, is it always a win for the freedom and power of the individual? Not unless you’re careful. Technology can be sustaining for totalitarianism as well as a disruptor. It can be a tool for surveillance, social control, databases of Muslims.

I like Uber, but a lot of the code is a pricing black box designed to maximize the value of Uber, not a free auction market. There are still questions of fairness to e.g. the disabled, blacks. You can use big data to understand customers better and offer them more and better choices, or selectively charge people more.

Build a true 2-way market where drivers can offer services and riders can bid for them transparently, maybe even based on a blockchain ledger, and that’s a true libertarian solution. If you work on free software, you are a true libertarian and a scholar and I salute you. Until then, you’re paying lip service to freedom while creating a new layer of code to regulate behavior, while maximizing your own rent-seeking.

It can sometimes look like tech hubris, liberty for me, the nerd, and not for thee, the unwashed mashes.

Since we’ve expanded the scope to theories of everything, what really separates left and right? Some interrelation between interests, learned values, psychological makeup:

  • How you feel about people making good choices on their own;
  • How you feel about authorities coming up with good policies and market designs, avoiding deadweight losses;
  • How you feel about markets, whether they are generally fair or some participants accumulate one-sided market power; whether externalities and market failures are important and common, relative to deadweight losses from interventions that interfere with markets like taxes, regs;
  • In particular whether labor market outcomes for individuals are fair; whether people generally get what they deserve in the market economy, or whether the distribution of wealth and income is unfair; whether people generally rise or fall to the level they deserve, or whether initial endowments of wealth, educational opportunity, ethnic biases are determining factors6;
  • How you feel about democracy and the likelihood of corruption and capture vs. the ability of government to work for common good;
  • How much slack we should give to the ‘losers’; how you feel about free riders;
  • How you feel about authority; are you oriented toward rules, and motivated by a sense of moral outrage, or outcomes, thinking moral rules are heuristics to lead to good outcomes; is your personality high in ‘conscientiousness’?
  • How you feel about the ‘other’; do you strongly identify with a group, are group values more important than individual values; is your personality high in ‘openness’ to the values of others?
  • Are you motivated by fear, or hope, is the world a dangerous, zero-sum place with internal and external threats, or do people generally work together for progress?

Decent people can solve problems using left and right solutions, just as programmers can solve problems with different languages. But there is good code and bad code. Indecent, extreme, and disagreeable people aren’t going to solve anything.

We’ve become a nation of architecture astronauts, spending our energy on ideological flame wars instead of shipping code that works.

There’s a fine line between being a critic and a contrarian, and being a hater tearing down the progress other people are trying to achieve. Once you get to the extreme of opposing all mandatory vaccinations, I would say you’ve left reason behind. Given what we understand about network effects, advocating freedom to engage in antisocial behavior without sanction seems like rejecting the categorical imperative and reason-based Western morality.

Elevating individual freedom to the highest and only moral good, and following that to its extreme conclusion on matters of asteroids and vaccinations, seems like bad code. In the extreme, it leads somewhere between indifference to the possibility bad choices may create human suffering and outright cruelty. Of course, gentle libertarian-oriented friends, I don’t ascribe cruelty to you, but I urge you ask yourself how far extremism in the name of liberty should be taken before it becomes a vice, or at best self-defeating.

Men of intemperate minds cannot be free. Their passions forge their fetters. — Edmund Burke

To suppose that any form of government will secure liberty or happiness without any virtue in the people, is a chimerical idea. — James Madison


1 Suppose individuals solve problems by freely agreeing on institutions like democratic governments with civil liberties and also taxes, police forces, prisons, some middle ground between non-violence and slavery? If people freely enter into a contract that imposes sanctions for violations, is it still freedom?

2 Yes, Virginia, people identify, and act with their tribe, nation, other institutions, not just as individuals. People have aspirations as individuals and also for their social unit. Have you ever been to a football game or other ‘sacred’ ritual where people dress, paint their face, eat and drink, sing, chant and play music in the proper manner to appease the universe and bring success to their in-group (and disaster to the out-group)? The original sin of communism and totalitarianism is not respecting the individual and private property, evaluating every individual choice as political and what it means for the state; the original sin of libertarianism is not respecting the drive to identify and find meaning through an in-group, ‘something bigger than me’, viewing every interaction as a trade, significant only to people as individual unitary atomistic me-bubbles.

Leftists tend to view market forces as somehow illegitimate when they are laws of human nature and game theory math. Libertarians tend to view everything as a market, and politics as somehow illegitimate. Markets and politics are just a couple of important, well-defined games of organizational and social control among many, such as social status.

3 If you think this is an easy problem, think again. Refactoring is a bitch. Some argue the 2nd Ave Subway costs more because it had to go deep below all existing infrastructure. If one entity owned all the buried infrastructure, you wouldn’t dig up the road one week to fix electricity and the next to fix water, you would presumably do everything in one shot, or redesign an integrated subway/water/gas/electric/telecom package and have something more maintainable that saved money. But then every time you need an electric fix, you have to go through the monolithic infrastructure agency, maybe wait until you can schedule water/gas/electric simultaneously. Sometimes you need agencies that cross jurisdictions, like the Port Authority which coordinates transport activities that impact the NYC area with NY and NJ. Maybe guns are the poster boy for this problem. You can’t regulate guns in Chicago if they are unregulated in Mike Pence’s Indiana. So I would prefer a national registration system, with local authorities in charge of local regs, but then at least they would be able to look up what residents bought in another state, require training, insurance, follow the trail of guns used in crime.

4 Proposition 13, which freezes California property taxes at 1975 levels, achieves the unusual feat of being rash direct democracy (variance), and locking in an outdated policy over the long term (bias). Premature sub-optimization.

5 Actually, that’s what machine learning is. But even in supervised learning, you define how the components can interact, and what they are trying to optimize and iteratively improve. Sci-Fi time! Imagine a society of the future where a giant computer is taught the human happiness ‘loss function’ and some kind of mother ‘hello Google’ and smartphone notifications directs everyone in the most optimal way.

6 Case in point: Civil rights. Was Jim Crow really more tolerable than the Civil Rights Act? Under normal circumstances a law targeting specific groups and specific outcomes, as opposed to universal principles, is bad code. But it’s perverse to deny blacks the right to participate in democracy, enforce a system of segregation by extralegal means and unequal application of the law, and then say that system is less of a cruel insult to freedom than a law that ends it. When that system was maintained in many cases by unequal application of equal laws such as bogus ‘literacy tests’, I fail to see an alternative to mandating reasonable outcomes as a last resort in this sort of extreme situation. It’s hard to do social science without thinking about outcomes; Noble principles and intentions should be eventually be checked against results. It’s called evidence-based policy. Ambrose Bierce said politics is a contest of interests, masquerading as a contest of principles. Better people acknowledge they are fighting for their interests (outcomes), but also think of others and shared principles. The worst sort actually believe they are fighting on principle, and only their opponents are fighting for their own narrow interests. Whatever your ideology, at the end of the day you have to ship code that works.