Bob Murphy is incredulous: Don't I realize that making an activity illegal tends to produce violence in those conducting that activity?
Yes, I do realize that, but the question is "Why does it do so?"
Let us look at some history to answer this question. The earliest form of human social organization was the band. No, not like the Beatles and the Rolling Stones, but a small group of closely related people, numbering perhaps a few dozen, who live, hunt, and forage together. Inside the band, the incidence of violence was low. When there was a dispute, the disputants brought their case to the band's elder(s). But violence between bands was widespread. Why? Here I forward a hypothesis: in a case of conflict between two different bands, there is no arbiter to whom they can bring their dispute for resolution, and so they fight it out.
As human population density grew higher, and this interband violence grew more frequent, a solution was devised: A number of bands in close geographical proximity bound themselves together into a tribe, with perhaps hundreds or a few thousand members. Violence within the tribe decreased dramatically, but intertribal violence was high. Why? Our initial hypothesis seems to hold good: within the tribe, disputants could bring their disputes to a final arbiter. But there was no such arbiter for disputes between tribes.
Tribes themselves consolidated into larger units, such as confederations or kingdoms. These were the forerunners of the modern state. Violence within these units decreased, but violence between them continued. Why? Once again, our initial hypothesis seems to hold. And many tribes recognized the benefits of this higher form of social organization: Germanic tribes often fought with Rome in order to gain admittance to the Roman Empire and reap the benefits of the Pax Romana.
As Murphy notes, the illegal status of, say, the trade in cocaine produces a great deal of violence. But why does it do so? To answer that question, we might ask, "In what situations this violence occur?" I think it is clear that most often, it is in cases such as a turf war between rival gangs, and not within the gangs themselves. Why would that be?
Our initial hypothesis seems to still be serving us fine: Within a gang, there is a final arbiter of disputes, namely, the head of the gang. But in disputes between gangs, there is no such final arbiter, and thus, disputes often turn violent. By making the cocaine trade illegal, the state has cast the various groups participating in that trade back into a pre-state condition.
And this explains quite well why various Mafia groups in 19th-century Sicily settled their disputes with violence: it is not because they were illegal, per se, but because they lacked a final arbiter for their disputes. In fact, Murphy has offered no evidence that under the legal code of the Kingdom of the Two Sicilies, it was illegal to offer private protection services. (And it would seem strange if it was: after all, private security is perfectly legal in the modern United States, and many other contemporary nations.) And even if it turns out that it was illegal to do so, what of it? The Mafia arose precisely because the Kingdom of the Two Sicilies was so lax in providing any law-enforcement in Sicily. The negative effect that the Kingdom had on the prevalence of violence between various Mafia families (and note the use of the term "family": here we have a return to earlier forms of kin-based social organization!) was that it would, no doubt, have acted to prevent any one family from consolidating rule over the island and forming a new state. An attempt to do that would have drawn the king's attention to the island, and have resulted in an invasion, no doubt. So here we have an instance of a state too weak to perform its job, of enforcing the law and providing final arbitration of disputes, but not so weak that a new state could form in its place. This is similar to the condition of many "failed states" today.