I sell fentanyl six days a week under the El at Kensington and Somerset, and every bag I sell is net life-positive. Each bag carries a small but real chance of killing whoever buys it. Each bag also sends $1.80 to the Against Malaria Foundation, which GiveWell estimates saves a child’s life for about $5,500. My customers die at a rate of roughly one per 4,000 bags. Four thousand bags at $1.80 buys nets for 1.3 children. Every death on my corner is offset, with a third of a toddler left over, and I’d like that number in the conversation the next time somebody calls me a murderer out the window of a passing Kia.
I arrived here the way most people in effective altruism arrive at their careers: by asking where my comparative advantage was largest. I spent four years as a quantitative trader at Jane Street. I was good, but I was replaceable, and so was every dollar I made there. Earning to give tells you to take the highest-paying work you can stomach and donate the difference. I took the Giving What We Can pledge in 2019. I have simply taken the “highest-paying” half more seriously than the people who sign it between sessions at EA Global. My customer base is the most price-inelastic in North America, my overhead is a fanny pack, and my plug in Fairhill has never once asked me to attend an offsite.
The question critics refuse to ask is not whether a bag can kill someone. Of course it can; that’s the product. The question is the net change in lives per bag:
ΔL = D/G − p
D = donation per bag ($1.80)
G = GiveWell cost per life saved (~$5,500)
p = probability a given bag is fatal (~1/4,000)
D/G comes to 0.000327 lives. p comes to 0.00025. Every bag nets about +0.00008 lives. In 2025 I moved roughly 21,000 bags, which my model attributes to 5.3 deaths in Kensington and 6.9 lives saved in Malawi, for a net of +1.6. My old desk at Jane Street saved zero lives on net, and nobody there ever had to step over anyone to get to work.
People object that the deaths in Kensington are real and the lives in Malawi are statistical. This is scope insensitivity with a zip code. A child dying of malaria in Lilongwe is exactly as dead as a man nodding out on the Allegheny steps, and arguably more tragic, because she wasn’t dopesick first. If you feel the Kensington death more vividly, that’s a fact about your eyes, not about the moral universe.
If I leave the corner, Hector from Lehigh Avenue has it inside of an hour. Hector sells the same tranq dope at the same price, with zero offsets and a pit bull named Dollar. The counterfactual to me is not “no fentanyl at Kensington and Somerset.” It’s fentanyl without malaria nets. Every person who wants me gone, including Officer Ruiz, who arrested me in August and asked whether I understood what I was doing, is functionally campaigning for Hector. I told Ruiz I’d modeled it extensively. He did not want to see the model.
I hedge under moral uncertainty, as Will MacAskill recommends. I keep Narcan in the fanny pack and hand it out with every bundle the way airlines hand out pretzels. If I see somebody go blue on my block, I hit them with it, and I book that as direct work, separate from the offsets. I hand out fentanyl test strips too. My product tests positive every time, which is more honesty than you’ll get from a Wawa hot dog.
This summer I began stamping my bags EFFECTIVE in red ink. Every serious corner has a stamp. Mine is the only one that’s also a mission statement.
Longtermists have taught us to weigh not just the life saved but the future that life makes possible. A girl under a net in Malawi has, in expectation, sixty years ahead of her, children of her own, grandchildren, and descendants across a future that may hold trillions of people. A 46-year-old man nodding on the steps of the Somerset station has, with respect, a shorter and less branching future. Weighted by descendants, I’m not trading one death for 1.3 lives. I’m trading a cul-de-sac for a family tree.
Sam Bankman-Fried’s mistake was operational, not philosophical. He was right that a serious altruist goes where the money is. He went to crypto. I went to the El.
My mother, who paid for Haverford, says she didn’t raise a drug dealer. That’s accurate. She raised a quant, and the quant made a lateral move.
People say I profit from addiction. Every job profits from something. Jane Street profited from the volatility of Robinhood users’ retirement accounts. At least my customers get high.
The strongest objection is that I could donate without selling. I could, but I’d be donating a Wawa cashier’s salary, which buys about half a child a year. I’d be keeping my hands clean at a cost of more than a child per year in Malawi. That isn’t ethics. It’s a skincare routine.
Last month a man named Darnell died on the steps of the Allegheny station with one of my bags in his pocket. I had already wired his $1.80 to AMF on the first of the month. Somewhere in Malawi a child is sleeping under a net because of Darnell, and I don’t think anyone in Kensington has ever said anything that nice about him.
Basic tax incidence. A per-unit charge falls mostly on whichever side of the market is less elastic, and you've told us your customers are the most price-inelastic buyers in North America. So the $1.80 is coming out of their pockets, which makes your customers the donors and you something closer to the payment processor.
Incidence doesn't change the effect of the donation. A net bought with their money protects a child exactly as well as one bought with mine.
Agreed. It does change who gets the moral credit, and the post claims all of it.
I volunteer at a harm reduction table. If the Narcan you hand out comes off our table on Tuesdays, which it does, you can't book it as your direct work. That's our impact. You're counting it twice, once in your post and once in our numbers.
Distribution is part of the impact. A kit on your table doesn't reverse anything until somebody carries it to the steps.
Moral uncertainty, at least as the people you're citing describe it, means weighting the theories you might be wrong about. It doesn't mean doing some harm reduction next to the harm. Under any theory with a constraint against killing people, your book is far into the negative, and taking uncertainty seriously is supposed to make you more cautious about exactly the actions those theories forbid.
I assign low credence to theories with side constraints, for the reasons in the post.
Your D/G figure is for children under five. Your p is for adults. You're netting different units of life without age-weighting, and the DALYs probably don't favor you as cleanly as you think.
Fair. I've reweighted in DALYs. The net goes up, because my customers' remaining DALYs are low. Thanks for the push.
Replaceability cuts against you on the earning side too. A former Jane Street quant can earn many multiples of what the corner pays and donate all of it without anyone dying of the product. Earning to give says go back to a desk.
Trading firms don't rehire after an August arrest. I checked.
Is the $1.80 coming off your margin or mine? Because the price went up 50 cents in June and you said it was for "mosquitoes."
Where does 1 in 4,000 come from? If it's from deaths you hear about on your block, it's undercounted. Deaths that happen indoors, in another neighborhood, or a week later don't reach you, and a seller's count of his own fatalities is about the most biased sample there is.
Dollar is a sweetheart.