"Key escrow" is an attempt to implement the cyberspatial analog of search.
Nope. "Key escrow" is far broader...
The result is certainly broader. Arguably too broad. I was simply trying to say that someone with the narrower motive of trying to implement warranted searches in cyberspace might reach for key escrow as a solution. Mainly for lack of a narrower mechansim. As I tried to say in paragraph (3), I don't think key escrow has to be mandatory to have some value (whether it's enough to make it worthwhile is the essence of the debate). Nor do I think there's any point in outlawing unbreakable cryptography --- your worst outlaws would use it anyway. Also, outlawing it would be more intrusive than required to implement warranted searches --- aren't there some relationships (doctor/patient, lawyer/client, priest/churchgoer) that the courts recognize as sacrosanct? The only arguable strategy, I think, would be for society to say "we're going to subsidize the escrowed key infrastructure so that it will be enough cheaper and more available that most criminals will opt for it for most usage" --- and rely on the power of human stupidity to make it pay off. This requires a comparison of the cost of that public subsidy against the law enforcement payoff (and a design for the distribution of who pays how much of that subsidy). And depends on being able to make a price and/or availability difference that's significant. And while Heinlein warns against underestimating the power of human stupidity, I must say I wonder how long we could expect that most criminals will remain insufficiently funded, educated, or motivated to avoid using the escrowed key infrastructure for incriminating activities. I'm not sure how to evaluate any of these.