What could bitcoins possibly have to do with sex trafficking? To answer that question, we must go down the rabbit hole to meet Rebecca Portnoff, a Research and Data Scientist at Thorn and a PhD candidate at UC Berkeley. In her own post on Thorn’s blog page, Portnoff mentions how she aims to “build tools and techniques that can empower investigators to crack open the recesses of the Dark Net, seeking out and finding the marginalized children who are exploited, abused and taken advantage of.” And that’s what led Portnoff and a team of other researchers to analyze the ugly recesses of sex trafficking businesses online using machine learning and blockchain data.
The problem: differentiating between legal prostitution and sex trafficking
When you see an online sex ad, how do you know if it is posted by a consenting sex worker or a sex trafficking ring? It is difficult for human beings, including the police, to mine the tens of thousands of sex ads that get posted every week to differentiate the legal from the illegal. To identify criminal activity, the police have to make connections between ads that may point to a single source that is posting the ads. But there are two main challenges to this:
- There is no framework to make such connections, and
- Analyzing such kind of data creates a negative psychological impact on the people involved in such investigations.
The analysis: machine learning based on stylometry
To address the challenges mentioned above, Portnoff and the research team used a machine learning algorithm based on stylometry. This algorithm can parse through data and analyze writing styles to identify similarities between the content contained within different ads. By doing this, the algorithm is able to say, with a certain level of confidence, which ads have been posted by the same source. Thus, any source that has posted too many sex classifieds can now be selected for further investigation.
The key: payments using bitcoin
While finding similarities in content simply tells us that the source is the same, it does not tell us who the source is. To crack this key element in busting sex trafficking groups, the researchers looked towards bitcoin. The thing is, websites such as Backpage, which is one of the world’s most popular repository for adult classifieds, are not allowed to accept Visa, Mastercard or Amex payments for adult ads. This makes bitcoin the sole choice of payment for these ads. For ad sources that had several adult classifieds under their belt, Portnoff and her team analyzed the publicly-available bitcoin blockchain to identify ownership in terms of the bitcoin wallet used.
From research to real life
Portnoff’s research paper, called “Backpage and Bitcoin: Uncovering Human Traffickers”, was presented at the conference on Knowledge Discovery and Data Mining in Canada in August 2017. Since then, it has received a lot of acclaim and the team is currently working with NGOs and police teams to get the tool up and running for real-life investigations.
Apart from the obvious social impact, this story teaches us one thing. Fraud detection is turning out to be an important use case of blockchain technology and is something that should be incorporated in your current landscape to ensure accountability. Having said that, the traceability of a transaction creator does raise some privacy concerns when used for external actors such as customers. This is something that needs to be further delved into. If you have any thoughts on the same, I’d love to hear them.
