There has been much comment lately among bloggers and others about the recent Supreme Court takings decision in Kelo v. City of New London, which is succinctly summarized here. The gist is that New London could use the takings power to grab private property that it then sold to developers under the rationale that this would promote economic development. There was no dispute that the city could have taken the property in order to build municipal court buldings
People on the right are mainly incensed with the decision while those on the left appear to be split. The views on both sides are presumably based mainly on policy preferences, as distinct from applying the rules of constitutional interpretation and parsing of precedent that judges are supposed to follow. This is fine with me for discussion's sake, even if one hopes that the judges go about it differently, as constitutional interpretation, however socially beneficial as a judicial practice, is intellectually on a par with the ancient Roman soothsayers' practice of interpreting animal entrails.
Leaving aside constitutional interpretation to focus on policy, the case presents a straightforward (if hard to resolve) political economy question. The reason for a mandatory takings power, which doesn't line up especially tightly with public vs. private use, is the holdout problem if there are multiple property owners. Thus, suppose a road would be worth $100 million to the society, that it requires knocking down 100 homes that are worth $100,000 each, and that each of the 100 homeowners wants to shoot for the entire social surplus of $90 million if they can't be forced to sell.
If we assume a benevolent government, that's pretty much the end of the story. If we instead, more realistically, assume concern about how the takings power will be used, but we still want to have some such power (as even Richard Epstein would agree), then the question becomes where to draw the line as a kind of filter between what are more likely to be good uses and what are more likely to be bad ones.
Public versus private use is certainly one possible way to draw the line, despite the arbitrariness of the distinction between direct public use and indirect public benefit from private use (i.e., positive externalities). Private use might be considered likely to be less strongly connected, on average in such cases, to valuable public goods provision that markets can't handle, notwithstanding the externalities problem that might arise in a given case. In addition, we might have a view that the private use setting invites misuse of political power. (E.g., George Steinbrenner uses his political clout in NYC to get cheap land for a new stadium via the takings power.) On the other hand, there could also be cases where well-intentioned governments find an opportunity to create social surplus by solving the holdout problem notwithstanding that the new use is being outsourced to be done by private parties (who might be better at it if cost-consciousness and responsiveness to consumer demand are important).
I therefore draw the weaselly conclusion that this is a tough question, to be decided based in large part on one's sense of the underlying empirics.