Your Algorithm Is a Zip Code You Can't See
· 7 min read · By Jacques Jean
There are two guys living in Austin. Same age, same neighborhood, same income bracket. Let's call them Marcus and Tyler.
They both open their phones on the bus.
Marcus sees a meme, a clip of a girl roasting her ex, an ad for a cash advance app, a video about how the system is rigged, another ad — this one for a job driving for a delivery service.
Tyler sees a clip from a podcast about angel investing, a short on prompt engineering, an ad for a circuit board prototyping service, a video essay about why the trades are undervalued, and an ad for a personal finance app charging him 4% APY on his savings.
They are in the same city. They are scrolling the same app. They are not in the same world.
I've been thinking about this a lot lately, because I run a hardware company and the kind of stuff I see on my feed every day — PCBA quotes, manufacturing tradeoffs, founders talking about supply chain — is not the kind of stuff most people see. And I didn't always see it either. The algorithm trained itself on me over years until my feed became a slow-drip MBA in whatever I was already curious about.
That's the part nobody really talks about. Your algorithm isn't just showing you what you like. It's quietly deciding what you'll know exists.
The receipts: algorithmic bias, measured
This isn't a vibe. There's data.
In 2019, a team out of Northeastern, USC, and Upturn ran ads on Facebook using completely neutral targeting — no race, no gender, no zip code filters. They just let the algorithm decide who saw what. The results were ugly:
- Job ads for the lumber industry went to an audience that was 72% white and 90% male.
- Job ads for supermarket cashier positions went to an audience that was 85% female.
- Job ads for taxi companies went to an audience that was 75% Black.
- Ads for homes to buy went to mostly white users. Ads for homes to rent went to mostly Black ones.
The advertisers didn't ask for any of that. The algorithm did it on its own, because it learned from years of click data that those were the patterns that maximized engagement. The DOJ sued Meta over the housing piece in 2022. Meta settled.
Then last year, researchers in China did something even more interesting. They set up 120 cloud phones — basically fake users — and simulated people from developed cities versus underdeveloped cities scrolling TikTok. Same behavior, just different "locations."
The users from developed cities got significantly more professional, expert-driven content. The ones from underdeveloped cities got less. The effect was statistically significant (p = .003 if you care about that kind of thing). And here's the kicker: when users from both regions actively asked for the same kind of content, the gap disappeared. Which means the algorithm was making the call by default, not because the users were behaving differently. It was guessing what they "should" see based on where it thought they were.
That's not personalization. That's pre-sorting.
Why your algorithm is the new zip code
For most of the twentieth century, where you grew up determined what you saw. If you lived in a neighborhood with hardware stores and small businesses, you grew up watching adults solve problems with their hands. If you grew up around lawyers and bankers, you absorbed a different vocabulary by osmosis. Your zip code shaped your imagination of what was possible.
Your algorithm does the same thing now, just with more precision and less accountability.
A kid who watches one video about flipping cars on Marketplace will spend the next month getting served 200 more. A kid who watches one video about FPGAs will get served the next 200 from a totally different universe. By the time they're 18, they've each absorbed thousands of hours of completely different "normal." Different role models. Different sense of what jobs exist. Different price tags on a normal life. Different vocabularies.
Here's the part that's hard to prove but feels right: this compounds. Slowly at first, then all at once.
The honest caveat
I'm not going to pretend the science is settled. It isn't.
The actual causal link between "your algorithm fed you X content" and "you earned Y less money over your lifetime" is genuinely hard to measure. The racial wealth gap in this country is overwhelmingly driven by stuff that has nothing to do with TikTok — inheritance, housing discrimination going back generations, unequal access to credit, the GI Bill leaving out Black veterans. Algorithms didn't invent inequality. They inherited it.
And some of the more dramatic claims about filter bubbles have been pushed back on. A big PNAS study last year found that short-term exposure to filter-bubble recommendations had pretty limited effects on people's actual opinions. So I'm not going to sit here and tell you the algorithm is a direct line to your bank account.
But here's what I think is happening. We're maybe 15 years into algorithmic recommendation at scale. The first generation of kids who grew up with their entire information diet curated from age six are just now entering the workforce. We won't really know what the effect of that is for another decade. The early signals are not great.
The contrarian play
If you take all of this seriously, you have two options.
The first is to fight the algorithm. Curate ruthlessly. Follow people who are smarter than you, in fields you don't understand, who will pull you into rooms you didn't know existed. Unfollow anyone who only confirms what you already think. Search for the stuff you should be seeing instead of letting it find you. This is hard work, and it never ends.
The second is simpler. Get off it.
Not forever. Not even for a month. But long enough to notice that the feeling of needing to check is not the same thing as needing to check. Long enough to remember that the people you actually live around — your coworkers, your family, the guy at the gas station — were once the people who shaped your sense of what was normal and possible. Long enough to read a book about something you didn't know you were curious about. Long enough to touch grass.
The research on quitting social media is mixed. Some studies show big well-being gains, others show none. But the studies aren't measuring the thing I actually care about, which is: did your time spent away change what kind of person you were trying to become?
That's the question. Not "am I happier without it." But "am I steering, or being steered?"
The algorithm has no opinion on what you do with your life. It doesn't care if you become a founder or a delivery driver. It doesn't know what would be good for you. It only knows what keeps you on the feed.
You're the only one who can run toward something. The algorithm will only ever pull you sideways.
If you liked this, send it to someone who needs to read it. And then close the app.
All posts · Home