This is likely a more erudite audience than I was envisioning when I wrote it...

As opposed to the morons who read Ribbon Farm? Kidding, obviously :)

More seriously, neither top down nor bottom up can lead to optimal solutions, but for vastly different reasons. Top down control we can reject out of hand for all but the simplest of systems, because it is impossible to define a set of regulations and rules that can account for all complexities and contingencies that are inherent to a society. Your three strikes example is illustrative in that it's ostensibly a simple rule made to have simple results: That is, rational actors will not continue to commit crime in light of the cost benefit trade off of marginal personal gains vs. a long prison sentence. It utterly fails to recognize that humans aren't Homo economicus or even something resembling such. Thus the effect of such laws is prison overcrowding, sentences that are draconian compared to the crime in many cases, and drug addicts being sent to prison for essentially harming themselves. The consequences are bleak and apparent, though no politician has the fortitude to do anything about it, due to the system stability that is our voting method (winner take all elections that serve to embolden the extreme and harm free expression).

Bottom up has a very different problem, which is that it is necessarily heavily dependent on historic contingency and initial conditions. For example, suppose a bird is born with a mutation that allows it to fly higher, faster, farther while expending less energy. This bird, according to the theory of survival of the fittest will pass on her genes to her progeny and so on. However, what if this bird were born in a year with a sufficiently harsh winter that her mother dies, and she starves before she can realize her potential? Her gains shall never be realized. Alternately, and perhaps more illustratively, she is the second born in a clutch of two, and the species has already evolved to only feed the first offspring, the second being a kind of biologic insurance policy (and this is a real thing in some bird species). The stability that is built into the system will kill her chances (literally) of effecting change.

We see the same thing in society. When the energy well is sufficiently deep, it is impossible to effect change in an optimal direction, because doing so would either be too disruptive to the current system, or the change isn't part of what Stuart Kauffman calls the adjacent possible, which is to say systems can only move from one state (I mean a thermodynamic state, to be clear) to a state that is very close to it (one can jump from two to three, or two back to one, but not from two to five, say). I suppose my point is that even when starting a system from scratch, we are still limited by the initial possibilities (which are finite), as well as by the initial/boundary conditions into which the system is born.

As an interesting aside, I think Obamacare may be a good example of a governmental attempt at growing a system, and how that can be completely derailed by entrenched special interests. The theory behind the ACA is actually a small government, conservative idea. The idea is that there are three "pillars" that must be intact to support the system. One, that coverage is available to everyone. Two, that coverage must be bought by everyone. And three, that subsidies are available on a means tested basis so that poor people aren't unnecessarily burdened. These three rules of the game are supposed to set up a competitive marketplace whose outcomes will be competition and the elimination of the free rider effect. All of the law could be written concisely and clearly. Instead, what we have is a multi-hundred page behemoth of a law that for many is a stand in for government waste, bloat, and cronyism, all extras ostensibly for 'cost control', while unsurprisingly benefiting the very folks who "helped" craft the law (insurance companies!). I have no idea if the ACA would have worked on its original premises alone, but forgetting whether it works or not, it's a great illustrative example of a push-pull between bottom up growth and top down control.

So in the end, although neither is optimal, I'll take bottom up the majority of the time, given that it allows for human ingenuity much moreso than does bureaucracy. mk is fond of saying that he made Hubski just to see what would happen. Give people the rules of the game, and then keep your hands off. I think it's working so far, but I try to remind him and others about that every time changes are being contrived: will this change limit or expand possibility? If it is limiting, I'm usually against it, maybe for no other reason than to be contrary.