Nozick compares and contrasts “design devices” with “filter devices” in the fourth and fifth subsections of Chapter 10 of ASU (pp. 312-320). (He also appears to take a parting shot at Rawls. More on that below.) Here, Nozick temporarily concedes, for the sake of argument, that there is a single universal utopia, that there is one kind of ideal society that is best for all men. If so, how would we go about discovering what this society is like? Nozick describes two methods of discovery: (1) a priori design and (2) ex post filtering.
The design method of discovery is a priori. First, a designer starts from scratch or uses his knowledge of the world to create or design a blueprint; then he builds something or make a discovery in accordance with his blueprint. In short, you design ahead of time the thing (whatever it is) you are trying to build or discover, or as Nozick notes (p. 313): “Design devices construct something … by some procedure which does not essentially involve constructing descriptions of others of its type. The result of the process is one object. In the case of societies, the result of the design process is a description of one society, obtained by people (or a person) sitting down and thinking about what the best society is. After deciding, they set about to pattern everything on this one model.”
Filters, by contrast, are ex post and evolutionary because they operate by an iterative process of elimination over time. You try a bunch of things, keep what works, discard the rest, and then repeat the process again and again. In the words of Nozick (p. 314): “Filter devices involve a process which eliminates (filters out) many from a large set of alternatives…. Filtering processes are especially appropriate for designers having limited knowledge who do not know precisely the nature of a desired end product.” Because filters generally require less global knowledge than design devices do (especially when designers are starting from scratch), Nozick concludes that some combination of filtering with experimentation is the most optimal way of finding utopia. In particular, Nozick argues (p. 316, end note omitted):
“The filtering process, the process of eliminating communities, that our framework involves is very simple: people try out living in various communities, and they leave or slightly modify the ones they don’t like (find defective). Some communities will be abandoned, others will struggle along, others will split, others will flourish, gain members, and be duplicated elsewhere. Each community must win and hold the voluntary adherence of its members. No pattern is imposed on everyone, and the result will be one pattern if and only if everyone voluntarily chooses to live in accordance with that pattern of community.”
Okay, fine. The irony, however, is that Nozick’s overall utopian framework is more of a design device than a filter. That is, although his framework creates a filter by which many different types of utopias can be tested and discarded, the framework itself is a product of Nozick’s design! But in the course of this discussion, Nozick also appears to take one last shot at his intellectual nemesis John Rawls. And what a shot it is! Nozick (perhaps unwittingly) ridicules Rawls’s influential “original position” thought experiment in just a few short sentences (pp. 313-314): “Sitting down at this late stage in history to dream up a description of the perfect society is not of course the same as starting from scratch…. It is helpful to imagine cavemen sitting together to think up what, for all time, will be the best possible society and then setting out to institute it. Do none of the reasons that make you smile at this apply to us?” (Or to Rawls’s original position thought experiment?) After all, aren’t the imaginary beings in Rawls’s original position like Nozick’s cavemen?