Maybe Jonathan Wilde already had my idea in mind when he wrote about competitive government, but it's not exactly clear, so I'll write it myself. He explains why seperation of powers within the federal government hasn't worked that well and says that the federal/state split has been the far more important division. I agree with him -- alas, the system has nearly fallen apart. The bolding below is my own, and highlights what I believe to be the key proposal (and perhaps the next evolutionary step in human government).
In contrast to Federal legislation, a law made by the state of Virginia mostly only affects Virginians. Similarly, a law made by the city of Richmond, VA mostly only affects Richmonders. Without the power of a higher level government to countermand it, the narrower the geographic monopoly, the fewer people the law affects. If the costs of relocation are insignificant and individuals can costlessly move from one locality to another, then law becomes a private good. It only affects those people who choose to live in a particular legal jurisdiction. They capture the entire benefit of moving to that particular location and suffer nearly all the costs of choosing poorly. Similarly, those states capture the total benefit of attracting people to live there and suffer the consequences of people moving to another state. Highly distributed monopolistic governments in a world of costless switching results in law as a private good. The effects of such law are not borne by everyone, only those who choose it and provide it.I wish Mr. Wilde had written about the idea more directly, but it appears that he's suggesting that multiple governments with identical powers could operate within the same geographical area, and that residents could decide which government they would belong to, which laws they would be subject to, and which bureaucracy would receive their tax dollars. Whether or not that's exactly what he intended to write, I think it's a great idea, and I've thought about it before and tried to determine how exactly such a system could work.
If instead of moving geographically, the costs of switching were made nearly zero by making living under a different government as simple as picking up a telephone or clicking a mouse, then specific laws would only affect those people who chose a particular government. In such a situation, good law would be a private good and bad law would be an undersupplied public good. Aggression would become expensive as it would be much harder to capture a monopoly of the market for law. It would be much more difficult for governments to grow and tyrannize their citizens because the costs of that tyranny would have to be paid for by their subscribers who could easily “move” to another government rather than be an unwilling source of funding.
No longer would voters be able to vote themselves largesse from the public treasury. No longer would special interests be able to narrowly benefit themselves at costs borne by the rest of us. No longer would there be a tendency for governments to grow with each passing year. Currently, fighting the growth of monopolistic governments is an uphill battle. It is only the American culture of liberty that kept the US government from becoming truly despotic during the last century as governments in the rest of the world slaughtered their own citizens in record numbers. With polycentric law, libertarians have both culture and economics on their side.
There are two primary obstacles I can forsee. The first is obvious: how would a nation with multiple (say, three) governments conduct foreign relations? Interestingly, the new Iraqi executive branch is set up with a similar difficulty in having to represent three distinct racial groups: the Shia, the Sunni, and the Kurds. Perhaps a trinary country could have a single executive council composed of an executive from each government. The executive council could vote based on the population in each of the three governments, sort of like a mini-parliamentary system. For example, executives representing more than 50% of the population would be required to ratify a treaty, and before any executive could do so his respective Senate would also have to consent (using an example based on the present American structure). This could certainly be complex, but it might be workable. (Funding diplomacy and the military might be tricky.)
The second problem would be dealing with internal conflict between persons who belonged to different governments. Which laws would govern the dispute? What if one of the governments allowed, say, murder, and one of its citizens kills a citizen of a government which prohibited murder? It could get confusing, but is there any reason why the various governments couldn't set up "extradition" treaties among themselves, if they desired? But what if the murder-is-ok government didn't want to enter into such a treaty? Then I suppose you'd eventually get open warfare, but I doubt it would come to that.
Anyway, it's a mostly fanciful idea; polycentric government sounds too complicated for most people to want to deal with, and probably too complicated to be stable. If such a system were instituted I imagine that the multiple governments would be unified by treaty rather quickly and that the walls would break down, just as they have between the states and the federal government.