Change the Incentives, Change the World

Jason Kuznicki makes what is in my view an excellent observation about libertarian activism. Moral messages can indeed be effective, and indeed were arguably crucial to important victories like the end of institutionalized slavery, but moral messages alone can only go so far. Only ~10-25% of people in the United States self-identify as libertarian, and recent events have certainly shown that it is foolish to rely on some kind of grand historical trend of people caring more and more about freedom just because some of us expect them to. Rather, it is human nature to be attracted to ideas that are perceived as high-status, and that are adopted by people, cultures, or organizations that appear successful, and so arguably one of the best ways to get certain moral values to be widely adopted, whether libertarian, egalitarian, humanitarian, or otherwise, is precisely to build systems and institutions that operate with these values in mind and succeed because of it.

Social systems engineering as a form of activism has a long history that stretches across the political spectrum (and, for that matter, political compass), including projects that are libertarian, egalitarian, communitarian, and often a mix of all three. Along with the spectrum of politics, there is also the spectrum of implementation strategies, with the main divide being “broad versus deep” - building institutions that produce smaller and more incremental changes, but do so among a larger group, versus establishing communities that go straight into implementing their ideology in a very deep and pure form, though often at the cost of attracting fewer adherents and all but seceding from the mainstream society. Community currencies, kibbutzim, seasteading, the GNU General Public License, Tor, the original vision of PayPal, the PGP Web of Trust, and many other projects were all in their various different forms a part of this tradition.

So far, “broad” has proven itself to be a far more successful strategy than “deep.” Projects that attempt radical change all at once fail for many reasons, of which three particularly big ones stand out. First, making entirely new social institutions that are radically different from the status quo, and in a way that is better than the status quo, is hard. Second, deep projects often fail to scale, as systems that work in sizes smaller than Dunbar’s number often fail to retain cohesion once their size grows past it. Third, radical change in small groups often fails to show its virtues even if those virtues are present, due to the simple fact that such isolated communities give up many of the network effect advantages of living in a way that is strongly connected to the mainstream society.

So how do you do broad? The basic formula is simple. First, pick an area of life that you want to change. Or if you do want to change several things at once, make it possible for people to adopt your solutions piecemeal. Then, find some way of doing it that’s better than what we have today. Implement it. Wait until either (i) people realize that this approach works well and start adopting it, or (ii) you realize that you were wrong and you move on to the next thing. In many ways, this sounds exactly like entrepreneurship. The main difference, however, is that whereas entrepreneurship is usually about designing better user interfaces, or building tools that are more comfortable and convenient for their users than the tools that came before, what we are talking about here has more to do with reforming underlying patterns of behavior, and especially the incentives, monetary, social, and otherwise, that drive how we interact.

In my view, significant very successful examples of what I would call broad social systems engineering include the following:

  • Online ratings
  • Crowdfunding à la Kickstarter
  • Various sharing economy projects, perhaps most notably Uber
  • Wikipedia
  • The open source software industry

All are examples of relatively more decentralized systems successfully replacing more centralized alternatives, usually due to a combination of idealists who see decentralized approaches as being intrinsically valuable and ordinary users realizing that, in those cases, they are simply the most efficient way to do what those applications set out to do.

Examples on the horizon include prediction markets, more holistic and comprehensive reputation systems, and, yes, the entire blockchain and cryptocurrency space. Dominant assurance contracts may perhaps emerge as another example. They represent a radical change from the way that public goods are funded today. The patterns of thinking involved in participating in a dominant assurance contract, and the ways that the incentives of all the participants interact, are very different from the kinds of systems that we are used to today. But with information technology becoming more and more powerful, and more and more accessible and widely adopted with each passing year, the space for experimentation in social systems design will only continue to grow.



Let us turn our attention to blockchains and cryptocurrencies. Technologically speaking, a blockchain is simply a decentralized network that has a common shared memory (“state”), and updates its state by processing inputs (“transactions”) coming in from the outside according to a set of prescribed rules (“the protocol,” or “the state transition function”). Particularly importantly, this is all done without any kind of central coordination or single point of potential failure.

This should be interesting to social systems engineering enthusiasts for several reasons:

  • Blockchains can effectively run arbitrary computer programs “in the sky” in a way where anyone can verify their correctness, and where trusting the system does not require trusting any kind of central operator or developer. This makes blockchain-based applications very easy to bootstrap and greatly cuts down the barriers to entry in getting users to trust the application.
  • Applications built on blockchains cannot be easily shut down.
  • Blockchain applications can operate globally, cutting across jurisdictions, and are theoretically accessible to anyone in the world who has a computer and the internet.
  • Access to a computer and the internet is also all it takes to build a blockchain application.

Global accessibility, cutting barriers to entry, censorship resistance - all values that libertarians should certainly approve of.

One of the main applications of blockchains is “cryptocurrency” - purely digital money that benefits from all of these properties. After simple cryptocurrency transactions, the next step is “smart contracts” - using digital money that can be directly controlled by programs that live on the blockchain. For example, one can use a smart contract to implement a dominant assurance contract. Users know that they will get refunded if the contract does not hit its goal, not because they trust the operator, but because there is no central operator they need to trust. If the participants do not fully trust the operator to actually use the money to implement the public good, one can use a smart contract to release a small amount of money at a time, where participants can vote at any time to turn off the tap.

Another great application is prediction markets; Gnosis and Augur are doing this on Ethereum already. A third is the private-good version of crowdfunding, where the participants get something in return, and the purpose of the contract is to overcome a coordination problem. One could even imagine this being used to solve coordination problems in the “deep” kind of social engineering: imagine crowdfunding a town in the mountains, or a seastead.

A fourth is financial services like exchanges; one can use cryptography together with blockchain-based systems to build financial systems that provably act according to the rules and cannot cheat their users. For two examples of this, consider “proof of solvency” and “exchange with decentralized custody,” which are new methods of distributed assurance and trust that cryptography has made possible. Social means of gaining trust, like filing forms and getting bureaucrats to confirm that none of the directors has a criminal record, can be at least partially replaced by  technological means of gaining trust - building systems that reduce the need to trust individual humans in the first place. Finally, there is also the concept of “decentralized autonomous organizations” (DAOs) - entities, which can be for-profit or nonprofit, that exist purely on the internet with no real-world footprint, and whose rules of association exist only as smart contracts on the blockchain.

Aside from programmable money, blockchains are a great environment for building other kinds of decentralized social institutions, including reputation and rating systems, “self-sovereign” identity management, marketplaces, as well as various other applications like supply chain tracking, anti-counterfeiting, and authenticated data publishing.


So where do we start? One possibility is to take your favorite idea for a social systems innovation, find some specific problem in the world to apply it to, and go ahead and build it. Another is to take these ideas and apply them to a clientele that is already very interested in participating in this kind of experimentation - that is to say, use these techniques to solve problems in the blockchain and cryptocurrency space itself.

Here are a few examples:

  • Build a rating agency that evaluates the quality of various digital assets and informs users about which ones are trustworthy and which ones are likely scams.
  • Use proof of solvency and decentralized custody techniques to make a safer cryptocurrency exchange that will not blow up the way MtGox did.
  • Build a smart contract auditing agency. Set up a repository of smart contract codebases, and give them ratings on security and efficiency.
  • Build a decentralized reputation system and use it for peer-to-peer cryptocurrency trading.

Perhaps one reason why these types of applications have not yet been developed is precisely the fact that many of them are public goods, and public goods are themselves the sort of thing that is very difficult to provide without large centralized organizations. Well, maybe you can do it with a dominant assurance contract.

The space is still a young one, and the possibilities are many; if designing voluntary institutions that use the power of advanced information technology and cryptography to solve public goods, coordination and information asymmetry problems is your cup of tea, then I would strongly encourage you to take a look.

Also from this issue

Lead Essay

  • Jason Kuznicki recommends what he calls “libertarian social engineering,” which builds institutions and practices for a freer society. Libertarians are well versed in moral arguments, he says, but moral arguments alone tend to generate only a short burst of enthusiasm before fading. Much better would be to get to work on building the practical - and voluntary - future that we all profess to desire.

Response Essays

  • Some institutions aren’t necessary to a free market as a matter of principle, but they make life in market-oriented society much better. Those of us who believe that markets are a key part of a well-ordered society should take the lead in developing these institutions. Not only might they bring about a more voluntary society, they will also improve the quality of our lives. Tabarrok cites several some surprising examples of institutional innovations to justify these claims.

  • Edward Stringham argues that there’s been a lot more real-world private governance than one might imagine at first. Historically, banks, payment processors, and financial markets have commonly created their own, self-enforcing rules, which were often highly effective. A persistent habit among statist thinkers is to assume the necessity of the state; libertarians will often counter a persistent tendency to speculate. Neither is quite warranted, says Stringham.

  • Tom W. Bell suggests three ideas that might help achieve a freer society. Together they form a general approach to privatizing state-supplied goods. He cautions that libertarian engineering will have limited capabilities under existing governments; to significantly change how law functions, and to move it in a libertarian direction, may require new systems entirely. Bell concludes with a sketch of three possible features they may contain: shared ownership, jointly held democracy between citizens and private owners, and open-source law.

  • Vitalik Buterin argues that the way to build a more decentralized world is to make that world rewarding. Most people aren’t libertarians, and few will change their minds about politics simply because libertarians urge them to. But when specific social systems make it relatively advantageous to engage in market behavior, and when people can trust one another in a context that does need the state, their actions will take on a libertarian character anyway, and the world itself will improve, both by our standards and by those of others.