It seems odd to be debating Randy in this way since, as he noted, his office is just one floor below mine. However, it is Saturday and we actually live a couple of miles from each other, making it costly to talk face-to-face. Furthermore, when we do talk face to face, I apparently do not communicate clearly, as it appears that even after our conversation, Randy did not understand the distinction I was making between contracting out and privatization. Let me try to clarify.
First, consider the idea of “government production” and recognize that the fact is that “the government” never actually “produces” anything, including force, without contracting with “private entities.” Even if a bureaucratic organization such as a Department of Corrections (or police force or army) produces coercion, the individuals who work within that bureaucracy are private entities working under contract, negotiated either as an individual or through a collective bargaining organization such a union. The individuals are not “owned” by the state unless they are slaves. They contract to provide their labor services because they expect to better off than they would be in an alternative job. The benefits of the bureaucratic job may take many forms, of course, including any pleasure received from helping to produce what a bureaucrat perceives to be the public interest (a perception that is likely to be colored by self interests), as well as a good living to support a family and/or attractive life style, job security, perhaps pleasure from being in a position of power and authority, and so on (some contracts to provide labor to produce a government service are entered into under duress, of course, as with a military draft, in which case, the relatively unattractive alternative is not a job but some sort of punishment).
Now consider so called contracting-out. The contracting out for prison services (unfortunately called privatization by most policy makers and many academics) actually refers to a particular type of contract in which some part of the bureaucratic decision-making hierarchy is replaced by a decision-making hierarchy that operates under a different set of incentives than those faced within the bureaucracy. Profit maximizing firms under contract may produce the service at lower cost but that may not be efficient in an allocative efficient sense (that is why I put quote marks around “efficient” in my previous remark – I probably should have used something like “cost-effective”). Other differences in incentives also may arise (e.g., abuses of power may be less likely if the private firm can be sued and the bureau cannot). Importantly, however, the service is still purchased (financed) by a government organization dominated by individuals with their own objectives (a dictator, elected representatives, or bureaucrats with discretionary power, all of whom are subject to political pressures from powerful special interests). That is, the service is still paid for by coercively collected taxes, after all, and allocated to uses determined through political processes with all of their public choice problems. Indeed, one of those public choice problems is that contacting firms become lobbyists who have incentives to demand more spending on prisons and perhaps more criminalization, as I have noted elsewhere. Of course, corrections unions have similar incentives, but as Randy said in conversation, for-profit firms may be more efficient at lobbying too.
The point is that production of prison services by for-profit firms providing those services to the state is not what I mean by privatization. Indeed, in the context of our interchange, Randy’s statement that “purely private prisons in the current state-run legal system would still end up increasing incarceration rates” does not make sense. These firms are not “purely private” when they operate in our current, state-run legal system. They still provide services to the state and are financed through taxes, because, under the incentives that currently exist (e.g., the focus on punishment rather than restitution) there is very little private demands for such services. I would argue that “privatization” should refer to situations characterized by private supply AND demand. Privatizing supply alone (which, as suggested above, always occurs, except that the incentive structure changes when the contract is with a for-profit firm rather than a union or individual laborers) can make things worse rather than better. As I suggested to Randy, if Hitler had contracted out the extermination of the Jews, I expect that more Jews would have been killed at lower costs, but surely this contracting out does not move us in the direction of a libertarian society. Therefore, I really do not see how my opposition to contracting out for prisons is in any way contradicts my statement that “By knowing where you would like to end up, you are likely to be able to make better marginal decisions along the road, even if the destination is never reached.”
Libertarians who support contracting out presumably do so because they want a more cost effective (and/or less abusive) government, presumably so the government will become smaller. If I was a limited-government libertarian I might agree (in fact, I once did ). I am not convinced that they get the smaller government that they want, at least in the long run (e.g., as Randy points out, contracting out creates incentives for a new and perhaps more effective group to lobby for more government taxation and control in order to increase their own demand; and as I noted, making government more cost effective reduces the incentives to look for alternative solutions to perceived problems), but more importantly, as someone who hopes to see government wither away in the long run, I do not want to make it more cost effective at delivering “services” that themselves should be eliminated. Randy is correct when he states that “in anarchy any prison services would be privately provided,”  but he is wrong when he continues “so Bruce’s opposition to contracting them out appears to contradict what I [Randy] quoted above.” In anarchy (and indeed, any private market) the good or service is being supplied in response to the demands of private individuals. If the demand is privatized so expenditures are voluntary, then the services provided are likely to be quite different than the services provided to the government and financed by coercively collected taxes, even if the service is supplied by a for-profit firm.
 For more discussion of this point, see Benson, B. L. “The Market for Force,” The Independent Review: A Journal of Political Economy, 11 (Winter 2007): 451-458.
 See Benson, B. L. “Do We Want the Production of Prison Services to be More “Efficient”?” in Changing the Guard: Private Prisons and the Control of Crime, Alexander Tabarrok ed., (Oakland, CA: Independent Institute, 2003), pages 163-216.
 See Benson, B. L. “Third Thoughts on Contracting Out,” Journal of Libertarian Studies 11 (Fall 1994): 44-78.
 The idea of prisons in an anarchical society may suggest to some that a coercive state must then exist, but to the degree that prisons might arise, they would be very different than modern prisons. See the last chapter of Benson, B. L. To Serve and Protect: Privatization and Community in Criminal Justice, New York: New York University Press, 1998.