Jacob Lyles has an [excellent](http://distributedrepublic.net/archives/2008/12/24/down-with-policy-libertarianism) [series](http://distributedrepublic.net/archives/2008/12/27/structuralism-2) of [posts](http://distributedrepublic.net/archives/2009/01/04/different-libertarianisms) over at Distributed Republic about structural libertarianism and policy libertarianism – the question of whether people should advocate better governmental policies, or entirely different forms of government. Bryan Caplan responds that it is [Policy All the Way Down](http://econlog.econlib.org/archives/2009/01/policy_all_the.html):
> At first glance, what Lyles calls the SLs seem a lot more realistic: To change policy, you’ve got to change institutions, right? Unfortunately, institutions themselves are a kind of policy. They arise because previous institutions create incentives for change, and endure because current institutions create incentives for stability. Or as we economists like to say, “Institutions are endogenous.”
> Suppose, for example, that the SL advocates more federalism in order to get more libertarian policies. Isn’t this more “realistic” than advocating drug legalization? Hardly. The level of federalism is low and stable for a reason – when there was more federalism, political actors have incentives to reduce it; now that’s low, political actors have little incentive to change it. Alas, it’s policy all the way down.
I have mixed feelings about this, and I think I’ve found a distinction that makes things clearer:
There are two contexts in which we can think about structure. One is in the abstract: starting with a clean slate, how do we make a good government? (Even better, how do we make a system for generating and refining a variety of good governments?). I think structure is much more important than policy in answering that question. In fact, I think policy is mainly important there just as a way to analyze and understand various structures. The distinction really matters – to understand the policy a government will generate you need to understand the structure and its incentives. It isn’t just policy all the way down.
The second context is reform or activism. How do we get the existing world to be more libertarian? It is here that Bryan’s point strikes home. The same incentives in our current structures that cause them to generate non-libertarian policy will cause them to resist structural reform. Public choice applies to structure, not just policy. It is policy all the way down.
My political theories take both these contexts into account, as Bryan’s [follow-up post](http://econlog.econlib.org/archives/2009/01/except_seastead.html) mentions. I admit the second point, but rather than concluding that structural reform is impossible, I conclude that it requires starting a brand new state. And I believe that seasteading is the most realistic way of doing that.