Illiberal anarchy

New Zealand libertarian duo, Eric Crampton and Brad Taylor, have written a paper exploring the concept of “illiberal anarchy”. They look at the consequences of “meddlesome” preferences in both anarchy and democracy, and conclude that strong meddlesome preferences will cause anarchy to be illiberal.

Crampton & Taylor (C&T) suggest that with strong meddlesome preferences democracy may work better, while with widely shared weak meddlesome preferences, anarchy will be better. They go on to suggest that the nature of anarchy will lead to strong meddlesome preferences (undermining the system) while the nature of democracy will lead to widely shared weak meddlesome preferences (once again undermining the system). Bugger.

I think they make a few mistakes regarding the anarchist situation.

Anarchist bullies

C&T point out that in anarchy a sufficiently rich person can screw with other people’s lives. The example they give is that a group of sufficiently rich anti-drug prudes could offer strong enough financial incentives to encourage people not to associate with drug communities… and this will impoverish and isolate those drug communities.

That is all true enough, and there are plenty of other scare stories you could make. Expanding on the above theme, a rich group could simply pay everybody not to be your friend unless you got a haircut. My favourite is that a rich person could buy all the land around your house and then forbid you to trespass. There are a thousand and one other examples of how a group of sufficiently rich bastards could mess with your life, so C&T certainly haven’t stumbled across anything new here.

Their first mistake is to assume that the above examples violate libertarian “freedom”.

Here, I take freedom to have the deontological meaning of voluntary actions. The above examples end badly, but there is no involuntary actions, so there is no violation of “freedom”. It seems that C&T are defining “freedom” as “something that ends well”, which is a consequentialist definition that is in conflict with the deontological definition.

It is not problem with people using a different nomenclature, but unless we are clear about meanings it is easy to talk staight past each other. The libertarian concept of “freedom” is generally the deontological meaning. That is not to say that the consequences don’t matter… but for sake of clarity it is better to use different words for different concepts. For outcomes, we generally use the language of “utility”.

Their second mistake is to assume their anti-drug example has ended badly. As readers may know, I’m a big fan of illegal drugs… but that doesn’t mean that the easy and open supply of drugs is necessarily a utility-maximising outcome.

In a free society, people must be allowed to use influence on each other… and that includes paying people to do things that they otherwise might not want to do. Indeed, we do this all the time. The whole concept of a job is one person (the employer) using money to influence the actions of another person (employee) to do what they otherwise don’t want to do. Nothing wrong with that.

If, instead of paying you to work, a rich person pays you not to use drugs, or pays you not to associate with certain people, or pays you to cut your hair, then that is all consistent with a free world… and the outcome from free exchanges between people are generally utility-enhancing (otherwise you wouldn’t have agreed).

Worse than the worst case scenario

The third issue is that C&T appear to drastically over-estimate the liklihood of their life-boat example. The costs involved in trying to influence an entire city, let alone a country or world, to not associate with certain people would be astronomical (unless everybody already hated you).

Indeed, even in the most pessimistic of thought-experiments, there would likely be a significant minority (at least) who would rebel against the super-rich prudes. The truely massive costs of maintaining mass manipulation would likely be larger than the “benefit” from being a prude… even for the most prudish of prudes. And the possibility that there would be a non-prude with some money is effectively 100% (unless you assume the prudes have all the resources, in which case a more scary thought-experiment is that they will not share their food & starve you to death).

I know C&T wanted to consider a “worst case” scenario, but if we’re going to consider the effectively impossible then we may as well just assume that humans will refuse to deal with each other, or that aliens invade and eat our brains. It is best to stick to scenarios that are vaguely possible.

But their life-boat example doesn’t need to come true to support their generalised point. It is almost certainly true that people will end up in communities with rules that they don’t like. However, contrary to C&T, this outcome is perfectly libertarian and is no cause for concern. Human interaction usually involves trade-offs and compromises, and this truth of life doesn’t change for anarchists. The only difference is that under anarchy nobody can impose a compromise without your consent.

Armish anarchists

The next problem with the C&T story is that they assume that anarchy will trend towards a strong meddlesome tendency. Actually… that part isn’t the problem. I agree an anarchist society would certainly allow meddlesome peolpe to come together and form meddlesome communities and groups, including groups that would likely be banned in a democracy.

However, there is no reason to believe that these groups would ever get together to coordinate a massive (and massively expensive) campaign of trying to influence everybody to be like them. Indeed, the diversity of these groups (white racists, black racists, asian racists, strict catholics, strict muslims, strict buddhists, etc) would make coordination effectively impossible. Further, it is unlikely that many groups would actually want to convert the world. We have seen many such groups through history, and most simply want the right to practice their own weird ways.

C&T correctly recognise that anarchy allows more meddlesome groups to form, and then incorrectly jump to the conclusion that they will have the desire and ability to impose these meddlesome preferences on non-members. Just because somebody joins an Armish community, that does not mean they want to pay an absurdly high price in an ultimately futile attempt to convert the world to Armish life.

Indeed, while anarchy would allow people to join highly controlled societies, the internal dynamic of a voluntary existence is to encourage relatively more tolerance. Ironically, this point was recently made in a great little article by Brad Taylor.


It’s good to see people seriously discussing anarchy, and Crampton & Taylor have highlighted an important point — anarchy will involve influence (good & bad), compromises, crazy old nutters with money and strange agendas, weird little groups of xenophobes and cooks, and the potential for bad outcomes.

But ultimately these are not critical problems with anarchy. Influence (good & bad) and compromises will always exist and are best dealt with in a free world, while bad outcomes are possible in every political system. And contrary to the C&T story, rich crazy old nutters and xenophobes aren’t a big risk in a free world.

31 thoughts on “Illiberal anarchy

  1. The challenge as I see it in a liberal democracy is how to stop the government from making stupid rules and over taxing the people.

    The challenge as I see it in an anarchy is how to stop the government from making stupid rules and over taxing the people.

  2. Thanks for the comments, John!

    I think you’ve misinterpreted our argument somewhat. I’m sure that’s mostly our fault: it’s still a working paper and we need to work on the exposition. I’ll try to respond more fully later, but I’ll make a couple of general points (with which I think Eric will agree, but I could be wrong):

    On Anarchist Bullies: We weren’t meaning to restrict our argument to “bad outcomes” which don’t violate the non-aggression axiom. Prudes could directly enforce their preferences by paying their protection agency to punish drug users in much the same way as police do today. This would be aggression and unlibertarian under any reasonable definition. We used the contractual leverage example because the way that could work is less obvious and more interesting from a social science point of view.

    On Worst-than-Worse and Amish Anarchists: Meddlesome preferences can be a problem in anarchy without any meddlesome group achieving much sway over society as a whole. In anarchy, both the definition and enforcement of rules is decentralized. If a small group of prudes take even small efforts to punish drug users, they are restricting freedom to some extent. If there are lots of such groups, there will be lots of coercion, whether or not it’s coordinated.

    In general: I’m an anarchist and see the costs of bigotry in anarchy as less than in democracy (I think Eric is a fence-sitter due to Cowen and Sutter’s cartelization critique – I don’t think he’d say our argument is a knock-down against anarchy). The point is that, contra some utopian anarchists, there is always going to be some coercion and the chance that anarchy could end up less liberal than democracy is positive.

  3. G’day Brad, and welcome.

    On anarchist bullies… I agree that a security force could decide to expand it’s role to enforcing their preferences on others. But in that case they have become a government. That isn’t an example of anarchy leading to a bad outcome… it is an example of anarchy being unstable and resorting to statism.

    You say that meddlesome people can be broadly problematic, but I don’t think you paper showed that. They have three options to enforce their preferences, either (1) an influence based approach (as outlined in your paper) would would be prohibitively expensive and unlikely to widely succeed anyway; (2) a violence-based approach… effectively launching a war for conquest to re-instate a government; or (3) a violence-based approach… resorting to small time crime.

    I accept that (2) is possible… but I didn’t think that was the purpose of your essay. I thought (and please correct me if I’m wrong) that you used the assumption that you were in the “sweet spot” where “the network of protection agencies is strong enough to punish rogue agents but not so strong as to be able to form a government-replicating cartel”. So you were looking at potential outcomes that would come from anarchy, assuming anarchy is stable.

    If (2) really was your intended story, then my rebuttal was the wrong rebuttal. Instead I should have written about the relative incentives for war, why anarchist organisations are less likely to go to war than statist organisations (because they must raise the money voluntarily) and why richer societies are less likely to go to war than poor societies (value of life is higher, benefit from war is lower, opportunity cost of war is higher). But that paper is for a different day.

    With option (3) I think this would be better explained in terms of “whether anarchy or deomcracy would have more crime”. An interesting question… though if this was your primary focus I think you missed a hell of a lot of points.

    One main point is that crime is less necessary because you are more likely to find a place that works as you want it to work (al capone can just go to the alcohol city, pablo escobar can go to the cocaine city, etc). So the benefit from crime would be lower and opportunity cost higher.

    Then there are issues of whether a competitive security service would be more efficient & effective, which would increase the cost of crime. The possibility that a competitive security service would be less corrupt would also be worth considering.

    Your point might be that the desire to commit crime would increase under anarchy… but that also seems suspect and there are forces working the other direction (such as outlined in your paper on freedom & tolerance). But more to the point, if the underlying desire is to “change other people’s behaviour” is seems that casual criminal violence is a very ineffective (high cost; low possibility of success) method, and so that strategy will likely die out. Historically, when people want to change opinions they either use over-whelming force (invasion) or persuasion.

  4. In the Science Fiction novel, “The probability Broach”, the author has the hero raise a similar point, and he is told that in the anarcho-capitalist paradise he is in, very few people manage to make exorbitant fortunes, because there are no patent or copyright laws, so that Gates-type fortunes from Microsoft can’t occur. Hence, wealth tends to be evenly distributed. Whilst some people do have unique talents (like having good voices, etc.) keeping a fortune would still be hard work.
    So if some rich types want to fritter their wealth away on bribes, they wouldn’t do it for long.

  5. I’ve commented more fully on my blog. But with respect to stuff at #4 I hadn’t read before commenting:

    – Any use of force by private defense agencies doesn’t mean that the agencies have become the state. The agencies, and in particular the network coordinating disputes among them, has to be strong enough to keep Mafia, Inc from being a popular protection service. Caplan argues for a sweet spot achieved in the space between the costs of enforcing a coordination norm and the costs of enforcing a cartel (where folks have incentive to defect). If the agency members, via side-payments from the meddlesome, are happy with the suppression of libertine agencies, then that is no more the state than is the network that keeps down Mafia, Inc. At least within the terms of the Caplan/Cowen structure.

    I think I’ve covered most of the rest in the blog post. I don’t think we were trying to argue “this necessarily happens under anarchy” but rather that there are reasonable portions of world space where it could happen, that these parts of world space may be more likely to obtain under anarchy, and that the case for “pushing the button” even in Caplan’s sweet spot might not be as clear cut as we might initially have thought.

  6. Pingback: Illiberal Anarchy « Brad Taylor’s Blog

  7. John – In the above article you say that democracy leads to weak meddling which undermines the system. I don’t think it is correct to say that it undermines the system. Democracy may not be liberal but it can be quite stable. All weak meddling does is create a situation that leads you to prefer anarchy (until the fact that anarchy leads to strong meddling is factored back in). Of course all this assumes that such conclusions about what causes meddling are correct.

  8. On the return to statism: I like to think of the argument in terms of what activities are to be counted as crimes. To get an orderly and liberal anarchy, we want murder and rape prevented by protection agencies (i.e. treated as crimes) but drugs and prostitution allowed.

    If murder and rape are not effectively prevented, we’ve got disorderly anarchy. If drugs and prostitution are effectively prevented, we’ve got illiberal anarchy. If we ignore the morality of these acts and focus on the incentives people will face in anarchy, there’s no reason for distinguishing between real crimes and those which only injure a prude’s delicate sense of decency. It all comes down to the willingness to enforce your preferences not to be murdered and have others not use drugs.

    If it’s possible to have murder effectively prevented by private protection agencies in anarchy, it’s also possible to have drug use effectively prevented.

  9. Terje — That is a big topic, perhaps for another day. I think democracy will inevitably result in people voting themselves more handouts than they are able to pay for, creating an eventual debt crisis. But it’s probably stable for another 50 years. Though this wasn’t the case being made by C&T, so I should have worded it differently.

    Eric — Welcome. Kia Ora. I’m not saying that any use of force by a security agency is a government. I’m staying setting up a rule against the will of the ruled and then having a monopoly on enforcing that rule makes a government. Non-sustained acts of violence/coercion would just be “crime” and you’d be in conflict with the approved “rule-enforcers”.

    If you have a group (the rich prudes) setting rules for the libertine community and then enforcing those rules with some hired guns… then there is a government. The prudes & others still may live in an anarchy on their own land, but they have set up a government to rule over the libertines.

    Brad — I’m not sure why you insist that drugs should be allowed in anarchy. Smoking cigarettes is currently “banned” in many houses, my parents “banned” the X-Files in their house (don’t ask), gold clubs often “ban” people who aren’t dressed properly. I’m sure drugs will be allowed somewhere so long as enough people want them… but the “no-drug” rule will also exist in some places. The cost of trying to make a minority opinion into a universal rule through persuasion would be unrealistically high.

    In contrast, I doubt a “can-kill” rule will work anywhere, with little need for persuasion. Not many people will join the “can-kill” community, and I suggest that institutional evolution will shrink those hypothetical communities further over time. 🙂

  10. John: I mean drugs may be effectively prohibited everywhere, not in specific communities. If prudes have sufficiently strong preferences against anyone using drugs anywhere (puritanism is, afterall, the sneaking suspicion that someone, somewhere might be having fun) they will attempt to punish drug-users even outside those communities. I’m not insisting that drug users can exist in any community, much less in any private home, but that those who want to use drugs can come together in their own communities and do it in the open without interference from others.

    Religious groups wanting to prevent blasphemy or homosexuality might be a more realistic example than drugs. Again, they don’t need to be entirely successful in order to substantially restrict individual freedom. If I can’t hold my boyfriend’s hand while walking down the street in my own gay-friendly community for fear of a undercover bigot from neighouring fundytown shooting us in an effort to please his God, I’m less free. The requirement to take efforts to keep the bigots out will also restrict freedom.

  11. Brad, you will never be free from such dangers. Whatever you are, or do, someone will be against it. Part of the human condition is to be able to be different, even if you don’t like how others differ from you. I suggest you learn how to shoot, and buy a gun. (Self-defence was a big part of the world of ‘The probability Broach’.)

  12. Brad — I know what you mean, and I think it nearly certainly wrong given the current demand for drugs.

    I think you are drastically underestimating the cost required to try to maintain an anti-drug bribery strategy, and even if you had near infinite funds it would still be unlikely to work because some people will just put up with the inconveniences you create. Probably myself included. 🙂

    In reality, the only way they could even get close to enforcing their preferences is by setting up a police state to closely monitor the libertine community. But even if they did that, it is unlikely to work fully. Singapore has some of the strictest drug laws in the world and I found ecstasy and LSD within one day.

    So the prudes would either have spent trillions of dollars (not many people love their bigotry that much) or launched an invasion & set up a government, all in the name of something guaranteed to fail. Obviously, in a free society people who make those sorts of decisions will be on the wrong side of social evolution and will become less common over time.

    I agree that crazy bigots will exist in an anarchy and will commit crimes, as with any system. Presumably, you will live in a community that has some sort of security arrangement.

    However, in an anarchy people are more easily able to join together with like-minded people and avoid those who annoy them. And as Becker has pointed out, irrational discrimination is economically inefficient and so social evolution in a free world will choose against irrational bigots.

    One thing missing from your analysis was the institutional evolution of what you called “sects”. I think you’ll find that they will have a strong incentive to evolve towards greater tolerance of other groups (just as you suggested for individuals), and the ones intolerant towards outsiders will tend to die out.

    So anarchy will lead to more “sects”, but also more benign “sects”.

  13. I like the idea of an anarchist “state” within a very minarchist State.

    Does this seem plausible? We know that voluntary taxation is rare, but possible and feasible.

  14. John – I don’t think anyone is ever going to be capable of wiping out drugs completely, but the use of force to make drug use more costly reduces freedom. Prohibition by government doesn’t make drugs completely unavailable, but it forces us to incur costs to avoid detection and run the risk of being put in jail. Same situation could hold in anarchy.

    I’m not sure that social evolution really would favour benign sects. The major lesson of the economics of religion is that sects will require their members to make costly signals of their commitment. The cost of enforcing bigoted preferences then becomes a feature, not a bug. Becker shows that discrimination will be less common in markets. I don’t see that generalizing to the current case when you need costly signalling to keep the sect going.

  15. Brad — I agree that criminal activity has a costs.

    On sects, you are confusing in-group rules with desire to enforce those rules on out-group people. I think you are correct that anarchy will sustain in-group bigots, but natural selection will chose against those who prefer out-group violence (ie crime).

  16. Why is out-group violence selected against? Sect-members need to do something costly to demonstrate their commitment. Why not beating up fags or dopefiends?

  17. You already explained why out-group tolerance would be chosen over time in your Independent Institute article (you did it for individuals, but the same dynamic exists for inter-group analysis).

    So did Gary Becker in his work on the costs of irrational bigotry.

    A group that is known to encourage a criminal element will find it more difficult to interact profitably with the rest of society (unless they succeed in imposing themselves as the government). It is a high-cost strategy. In contrast, a group that fosters a cooperate stance with outsiders will get a clear economic benefit. Over time, the strategy that makes you money will win out over the strategy that costs you money.

    In your article to make the case that people will be more likely to join clubs. I think that is certainly true. But you provide no reason for thinking these groups will increase in their anti-outsider bigotry.

    An appropriate case study is the growth in membership of “friendly societies” during the 19th century. In the UK (where we have the best data) there was about 20% coverage in 1801… but that grew to about 55% coverage before they were killed off by the introduction of the welfare state in the early 20th century. So this is a good example of growing community involvement.

    These groups prided themselves on their tolerance, and while they had a range of (sometimes strict) rules for members, they unsurprisingly did not spend much time & effort on pursuing high-cost criminality. The suggestion that greater community involvement leads to crime waves seems to fly in the face of all theory and evidence.

  18. Generally speaking, anarchy will make most individuals more tolerant. What we’re saying in the illiberal anarchy paper is that it will also lead to a few people having very meddlesome preferences, and those groups that do will be more able to overcome the collective action problems involved in enforcing their preferences. Sects, as Iannccone explains here differ from what he calls churches (friendly societies, etc would fit in there) in that they require costly signalling by their memebrs, often by making themselves stigmatised by outsiders. When you need to show that you’re committed to the group and want to decrease your opportunities outside the sect, the standard incentive arguments against bigotry largely disappear. Killing infidels seems like a really good way to signal commitment and reduce your outside options. That’s why we expect to see more really bigoted people in anarchy than democracy. Further, due to their greater cooperative efficacy, they’ll be disproportionately awesome at enforcing their preferences.

    I agree that if we consider all the factors which influence tolerance in anarchy versus democracy, anarchy will come out ahead. We’re just pointing to one mechanism which goes the other way. There are going to be a lot less mildly bigoted people in anarchy, which I suspect is more important.

  19. The lessons of history are the obvious counters.

    Every corner of every continent in the world started out as an anarchistic society. Some where more successful than others but ALL of them eventually succumbed to statism.

    Statism is the natural state of being because most people prefer to have their decisions made for them.

  20. Saddly I agree with Yobbo. My instincts suggest that the best system that is both possible and long term stable is going to be a constitutional liberal democracy with some significant inbuilt biases towards liberty. The founders of the USA did a really good job but obviously not good enough.

    Additional ways to introduce a more liberal bias into a constitution include; mandatory sunset clauses on all legislation unless it satisfies a super majority. Term limits. A house of legislative review appointed by lottery instead of elections. Caps on per capita tax revenue that can be temporarily increased only via referendum (I’d say a timeframe equal in length to two terms of government would be the definition of temporary as few serious wars will go that long). Prohibition on governments issuing currency or bearer bonds of low denomination. Gold as the sole unit of account for tax purposes and government bonds and no changing the unit of account without a parliamentary super majority and a referendum. And obviously the usual structures and methods such as a federalist system and guaranteed basic rights (free speech, association, movement, trial by jury etc), recall elections etc. Citizen intiated referendums to recall (but not initiate) legislation should also be allowed for. I’d also go for a beefed up version of the charter city concept allowed for in the Californian constitution and ensure a clear method by which citizens could split their state. A clear pathway and method for states to secede should also be set out.

  21. Yobbo – There are historical examples where anarchy has worked fairly well. David Friedman’s study of medieval Iceland is probably the best place to start.

    Of course, none has lasted for more a few hundred years, but neither has any democracy so far. Democracy was generally viewed as an unworkable utopian idea before it was tried, too. Anarchy has never been tried in an industrialized society, and I don’t think we can just read the history books to know whether it’s going to succeed of fail. I don’t think there’s any general tendency in human nature to want our decisions made for us. There’s a tendency to be communitarian, which is fine at a voluntary level but causes problems when it gets expressed through government.

  22. Terje/Yobbo… I agree that anarchy seems unlikely, and historically hasn’t been too successful. There are a few notable exceptions (Medieval Iceland, Wild West USA) but not many.

    However, the same can be said about constitutional democracy. It has been tried more often, and failed more often. I think it’s an easier experiment to try, but I think it is more unstable than anarchy because the “democracy” part of the equation soon overtakes the “constitutional” part… as seen in the western world today.

    However, as I said above, I think the dynamic is changing. I suggest that the value of violence decreases as societies get richer… and as the value for violence decreases the viability of anarchy increases. If/when the value of violence gets low enough, anarchy will become a relatively stable and achievable option. The only problem will be getting there.

  23. Brad — I don’t think your paper showed an increase tendency to have meddlesome preferences. I think you showed that such people would more easily come together in groups, but I don’t think you showed a dynamic for the amount of out-group meddlesomeness (what a word) to increase.

    I certainly don’t think you showed that these meddlesome groups would be able to successfully spread their preferences. Do you really think the Waco mob would have been able to spread their ideas to LA if only they were left alone? Do you think the KKK would be more or less likely to have meddlesome preferences if they already had a KKK-land where they could safely escape those nasty jews & blacks (which is, after all, what they are asking for)? The friendly societies were always far more popular than weird sects, and had more influence.

    I accept your argument about signals, however the point of social evolution is that the signal that achieves its goal with the lowest cost to the group will become more popular.

    I accept that idea that some may actually cultivate an “anti-outsider” approach (and so the standard anti-bigot arguments don’t apply)… but this makes it even more certain that they will not be able to convert the outside world. They will be 100 nutters in a compound wearing nikes and shouting “the jews did it” to an outside world that looks on with scorn.

    I don’t think you can have it both ways. The behaviour in an anti-outsider sect is what guarantees they wont take over the world. Conversely, what allows them to have more influence in the outside world will make them more tolerant. Consequently, in a free world you would expect them to exist, but have little influence.

    Also, these “anti-outsider” bigot groups would have less chance of survival than a libertine community, as the libertine community wouldn’t be taking on extra costs, nor would they be attacking outsiders. This seems to put you in a catch-22. 🙂

    And also, I accept your argument that sects have lower internal transaction costs (as do friendly societies, churches, etc)… but in a free society they also have higher external transaction costs because they must deal with a thousands different outside groups.

    I agree with your summary position that anarchy will lead to a lower general level of bigotry, but will have small groups with high levels of bigotry.

  24. John – have you given any thought as to how a democracy might peacefully transition to an anarchy?

  25. I’ve been thinking a bit lately about future political trends, and I’m not optimistic about the direction of reform. I don’t think social democracy can reform into liberal democracy, let alone anarchy. And I don’t think social democracy is sustainable in the long term.

  26. I think anarchy would mean a great deal more democracy, Terje. All government property (eg Bruce Highway) would be privatised with everyone in the relevant town/state/country getting an equal number of shares which can be sold. Voting for the boards of directors could be largely automated by proxy votes (I might nominate John to vote for me in most elections, for example).

    I think a social democracy can be reformed into a liberal democracy. People just need to vote for the LDP!

  27. TerjeP, I’m not sure that it can transit peacefully. Probably secession is the only way. Britain went through Cromwell’s war, and the ‘Glorious’ Revolution of 1688, before Parliament became supreme in the UK. Struggle and strife might add strength to any new social order.

  28. Actually, TerjeP, an independent Tasmania might be what we should aim for. It has defined borders, costs the Feds more in subsidies than taxes pay for (so they might be relieved if it wasn’t their problem), and could do with a purpose to galvanise the place. All we have to do is settle there in large numbers, arm everyone against the menace of the Tasmanian Devil (so they all have guns), and win government- and then declare our independence! On paper, it seems easy!

  29. Pingback: Private Policing I Find Disturbing « Brad Taylor’s Blog

Comments are closed.