Terbium Labs is a data intelligence firm that says it specializes in the dark web, and claims that its produced the first legitimate analysis of the Tor network. The study -- which scraped a selection of sites -- reports that more than half of all traffic is legitimate, and that the amount of truly scary stuff is negligible. Much like that teenager's bedroom, there are lot of drugs and a significant amount of porn, but you don't need to worry about calling the FBI just yet.
For instance, the study claims that porn makes up just 6.8 percent of all Tor traffic. Illegal pornography (listed under the umbrella term of "exploitation") accounted for one percent of traffic. That, unfortunately, is still one percent too much. Drugs, meanwhile, cover 12.3 percent of all traffic, while non-prescription pharmaceutical sales makes up another 3.2 percent. For the purposes of the study, marijuana was included as a drug given its inconsistent definition across the United States.
There's often talk about how shadowy bodies use the dark web to buy and sell weapons of mass destruction, but the researchers didn't find much evidence of that. In fact, the pair only found one example of extremism and precisely zero weapons sales sites. This conclusion, however, could be an indictment of the methodology used to prepare this study, which may not reflect the breadth and depth of the Tor network.
The team only sampled 400 URLs that Terbium Labs' crawler accessed on a single day (August 5th, 2016) and there are several caveats. For instance, researchers blacklisted URLs that they believed to include illegal materials and didn't analyze them directly. Those sites were still included in the data selections, so the conclusions it draws may not be entirely reliable. In fact, a page in the report even ends with the slightly blasé point that "Research is hard."
That said, it appears that more rigorous analysis of the dark web is revealing that it's far less intimidating than it was believed to be. A study from 2015 found that there were just 7,100 .onion sites available to crawl, meaning that one scan took less than three hours.
It's certainly not conclusive, and there are plenty of reasons to want to inspect this data far more thoroughly. But it does look as if the notion of the dark web as this big, unassailable beast lurking in the bowels of the internet doesn't match reality.