Modern communication networks, such as cell phone systems and the Internet, have provided researchers with the opportunity to study human associations and movement on a much greater scale than previously possible. Almost all of the papers that describe this sort of network analysis notes that it could have real world applications, since existing and emerging disease threats can spread through social and transit networks. A paper that will be released later this week by PNAS, however, skips the whole "this may be a useful model" aspect, and goes straight to a network in which diseases actually do spread: prostitutes and their clients.
Although organized prostitution is apparently illegal in Brazil, there are no laws against receiving payment for sex, making it possible for sex workers to freelance. Like everything else these days, that trade has found its way onto the Internet, and some enterprising Brazilians created an ad-supported public forum for individuals on both sides of the transaction. The forum is heavily moderated to keep it strictly on-topic: sellers (aka prostitutes) can advertise their business, and those that partake can rate the experience, as well as provide some information about the precise services rendered (the focus was strictly on heterosexual prostitution in this system).
This rich repository of data turned out to be a fantastic opportunity for Swedish sociologist Fredrik Liljeros, who obtained six years of data—the entire history of the forum—and teamed up with two members of physics departments in order to perform a network analysis on the material. Overall, 6,600 prostitutes and over 10,000 clients were included in the study.
Crunching the numbers showed that the number of people networked to a typical seller was beginning to flatten out, meaning that many prostitutes in the forum were seeing about as many clients as they ever would. This, to the authors, suggested that the average time in the business might not be too much longer than the study period's six years. That wasn't true for their clients, leading the researchers to suggest that "sex buyers stay in the commercial-sex arena longer than do sex sellers."
Ratings did, apparently, make a difference, as the most highly rated sellers did in fact attract more business. There was little difference in the career trajectories between those who received low and moderate ratings, though.
From the perspective of potential disease spread, there was some good and bad news. On the bad news side, the longer that a prostitute was in the system, the more likely they were to engage in activities that are considered high-risk when it comes to pathogens, such as performing oral sex without a condom.
But the networks within the system were often small and fragmented; the authors say that the number of small networks is far larger than previous studies had reported for online and offline dating networks, for example. Although there are some very large networks within the forums, the trend towards fragmentation should slow the spread of diseases.
Many of the individual networks were clustered within urban centers. There were only about ten cities represented in the data—as the authors put it, "the data put rather large error bars on the results"—but it appears that the number buyers of sexual services scales roughly linearly with a city's population. That's not true for the number of prostitutes, however, as there are fewer than you'd expect from a linear relationship.
The vast majority of transactions also took place between people who live in the same cities. This means that diseases might spread rapidly within a single urban center, but having it spread between cities will be a less common event (at least via prostitution). That means that identifying those individuals who do mix travel and prostitution may have a disproportionately large impact on disease control.
It's easy to focus on the prurient aspects of the study, but most of what we think we know about the spread of diseases in human social networks is based on networks, like those of cell phone users, that get used as models for real world contact. In this case, the authors got a glimpse into a real social network of the sort that we know can act as a vector for disease spread and, as they point out in the discussion, there are significant differences between what they found and previous reports of social networks.
Source: ars technica