This article originally appeared on THUMP Germany.
Have you ever shared stories about drugs or recollections from a chaotic party with your friends on Facebook? No? Perhaps you're the cautious type. That's smart, but it doesn't matter as much as you might think. A recent study from the University of California at Berkeley found that, thanks to a new algorithm, it's possible to determine whether you consume drugs or alcohol by analyzing your likes and status updates, German digital rights advocate Netzpolitik.org reports.
Sounds like something out of George Orwell's 1984, doesn't it? The study was supposed to be used in tandem with drug prevention research. But let's start at the beginning.
The USC Berkeley study appeared last month with the title, "Social Media-Based Substance Use Prediction." It analyzed 11 million Facebook users, and an additional 22 million status updates from 150,000 people. The goal of the study was to identify people who suffer from substance abuse disorder, or who could be prone to it.
The researchers developed a machine-learning software that analyzed certain data sets—in this case, combinations of keywords sourced from other studies of people who suffered from substance abuse disorder—to find correlations, patterns, and predictions for the future.
The keywords include curses, statements about physical and mental states, and words with additional sexual connotations. Music, film, or entertainment-related preferences were also included in the study. Someone who likes animated movies is less likely to consume alcohol in excess, but if you're a fan of V For Vendetta, you might be at greater risk, for whatever reason.
The study showed that the machine-learning software has a high capacity for prediction: it was between 80 to 86 percent accurate when identifying people who suffer from substance abuse, including tobacco, alcohol, and general drug addiction.
That level of accuracy greatly exceeds previous studies, which has led the researchers to the conclusion that social media is "a promising platform for analyzing substance abuse disorders, [in that it] makes drug use more visible and… facilitates prevention."
However, it's not exactly clear how the findings of this study could help prevent drug abuse. Should someone contact a Facebook user directly if it turns out they're at risk for substance abuse disorder? And if so, what exactly should they be told?
Prevention is an obvious part of drug policy, but many experts argue it's more important to decriminalize drugs, as well as advocate for safer use and harm reduction.
Ultimately, the data collected by the study is much more interesting for other groups of people—especially businesses that advertise on Facebook, or governmental authorities looking for offenders who've violated federal narcotics laws.
Translated by Olga Mecking