Yeah, they do — but there’s a critical flaw with their method of data collection that renders any “official” information on black-market operations at least mildly flawed, assuming the data has any degree of accuracy at all.
The response bias is one of the first experimental constraints taught in statistics classes, yet, the government doesn’t attempt to control it at all.
To put it simply, a census is a great way to gather information about regular people — the kind that pay their taxes, get their expired IDs replaced, and have fear of fines and legal complications imposed by not complying with the government.
But drug dealers, on the other hand, clearly don’t even fear prison, let alone fear getting caught, so they probably aren’t scared of the government’s threatening reminder mail to fill out a survey.
Another flaw with published facts on drug dealing and usage habits is the timing of the data collection.
Besides the government, a lot of information in this domain is collected at places like hospitals, rehabilitation clinics, and therapy — where it is often either too late in the drug patron’s story to gather anything of use, or it’s a situation where they have a strong incentive to lie.
Imagine you were a heroin addict without a lot of money, being asked by a doctor to assess how extreme your symptoms were.
If you were honest, you could be forced to shell out thousands of dollars of money you don’t even have, for extra treatment and care.
Alternatively, you could downplay your drug addiction and avoid that potential financial havoc.
So as one could see, Venkatesh’s data collection method, while perilous, was undeniably more reliable than any government survey could be.
Venkatesh befriended one drug lord in particular, whom he dubs JT, that I found particularly interesting.
JT was unlike the picture of black-market America painted by the media — he owned his home, had a bachelor’s degree, a family, and a previous job in corporate America.
This screams anything but Chicago poverty.
JT was a seemingly normal guy, who turned to dealing drugs simply because it made him more money than his cubicle job.
“Managerial prowess” and “entrepreneurial skill” are two ways Venkatesh describes him, and perhaps these descriptors are rightfully so — JT manages the Black Kings, a prominent drug-dealing, trading, and organized crime franchise on the streets of Chicago’s historically impoverished South Side.
Venkatesh’s experiment proves that the method of data collection through direct observation reveals truths that don’t come out of the census, academic research, or surveys.
Once you’ve eliminated the response bias, you’ve opened yourself up to a whole new realm of information that your subject’s lying by omission was previously covering up.
I wanted to also repeat some version of this experiment, minus the risk of arrest and death, to study the drug dealing habits of my own peers.
My first thought was towards social media — an excellent breeding ground for this sort of information, where teenagers and young adults open up about themselves and obscenely overshare useful information, sometimes to their own demise.
One app in particular, though, turned out to be more useful for me to look at drug dealer activity without a response bias.
It wasn’t people’s Snapchat stories of them partying, or their “finsta” posts of the morning after, where I found the greatest wealth of information.
In fact, it was on a platform that nobody even considers to be social media.
I found the information I needed on Venmo.
Venmo is the payment app for young people — it’s crushed all of its competition.
Even banks rolling out their own pay-back apps haven’t been able to match the ubiquity and market performance of Venmo.
The company, which was acquired by PayPal, has become a household name among college students, who use it to pay for everything from rent and utilities, to event tickets and food, and of course, drugs and alcohol.
To understand why Venmo is being used so widely for illicit sales, we first need to go back to before, when college students needed to pay for “the goods” in cash.
No underage person would have been able to pay for their cigarettes (before the rise of the Juul) or beer with a credit card, since that sort of possession utility just did not exist before mobile payments.
Cash was king for these transactions, since it provided secrecy and convenience, which are two critical components for these small pseudo-crimes occurring probably hundreds of times a day on college campuses.
Venmo also provides these same two convenience factors.
A student in need of some substance could probably message their supplier (likely another student), then pay for their substance on the spot using the app.
In terms of anonymity, they can choose to hide their payment from the public “friends” feed in the app, but this arguably isn’t as important to dealers and buyers, since nobody really scrolls down their Venmo feed like they would their Facebook timeline, or Instagram feed.
More importantly, the payments aren’t identified or itemized on bank account statements, meaning that user/dealer parents aren’t able to know what their children are purchasing on Venmo, even if they’re looking at account statements every month.
So naturally, Venmo has replaced cash as the king — well, at least when it comes to illicit transactions among college students.
What sorts of questions can you answer from a public Venmo feed?.Well, for one, you can’t really answer anything about drug usage habits, because you don’t know who’s actually consuming the drugs, and how often.
Dealer pages are often identified by certain slang terms that are easily identifiable to a college student who uses those terms, but probably too nuanced for any future NLP-drug-deal-prevention algorithm to pick up.
Thinking forward in terms of the language processing I’m going to use for this experiment, I’ve compiled a list of red flags indicating someone is a substance dealer, meaning a regular distributor of anything from birthday champagne to adderall to Juul pods, from their public Venmo profile.
These characteristics are through personal observation of individuals who have admitted to me or to others that they’re a drug dealer.
Drug dealers often identify themselves as “the plug” to their peers.
This comes in the form of usually an aptly-placed plug emoji in their Instagram bio, or acknowledgement of the dealer’s “plug” status in the public payment messages on Venmo from their customers.
Phrases like “thanks for the plug” or “shoutout to my plug” are like the drug user’s positive reviews on the dealer’s pages.
The transaction messages are normally also a great indicator of what sorts of payments are being made.
A prime example of this is the usage of the word “broccoli” to mean weed.
Sure, “broccoli” one or two times could just be some healthy roommates paying each other back for groceries, but when multiple different people are paying someone for “broccoli”, one would begin to question if this person really loves those cruciferous vegetables.
I have stressed this so much in previous research, and other natural language work, but emojis are perhaps the single greatest asset to any natural language analysis, especially online, ever since Stanford invented the NLTK.
Multiple people paying someone with the same popped bottle emoji at the same time is a clear indicator they’re paying them back for alcohol.
Some other obvious ones are the aforementioned broccoli and plug emojis, the puff emoji, and all of the ones that look like green grass or leaves.
So, I decided to put this into action by scraping and analyzing the Venmo profiles of some known college drug dealers in my network.
Check out the project here.
.. More details