Am I One of the Bad Guys?
Last week, Meta agreed to pay $17 billion to settle claims that it designed Facebook and Instagram to be intentionally addictive to teenagers. I was part of building these systems.
I think about the employees of Philip Morris. And of Elizabeth Holmes. Was there a moment when the recognition hit? Or is "making the world a better place" a slope so gradual you don't feel yourself sliding until you're already at the bottom, too far down to climb back? What does it take to come to terms with being one of the bad guys?
The (Blind) Optimist / Futurist
I am an optimist and futurist. For years I worked at design firms where Fortune 100 companies hired us to paint beautiful pictures of the future — what a connected home might look like in ten years, or the possibilities that data-gathering wearables might unlock for us. The work is close to what a science fiction writer does, except the story arrives as a concept video with realistic interface specifics that help with envisioning.
Now I coach people who feel stuck. I help them get clear on what actually matters to them, then build a roadmap of micro-experiments that gets them there.
The magical thinking is the gift and the danger. The same optimism that lets me paint a beautiful future for a boardroom is what makes it so easy to miss the harm a beautiful future can cause along the way. I wasn't naive at Facebook. I was doing exactly what I'd always done: seeing the future worth building toward, and trusting the rest would sort itself out.
My Golden Age of Facebook: 2015-2019
I worked at Facebook from 2015 to 2019. Some people call that stretch one of the golden ages of tech, and from the inside, it felt exactly like that. A wifi-equipped bus picked me up in San Francisco and dropped me in Menlo Park. I could pick between my favorite lunches of BBQ, steak salad, or tikka masala on a bed of naan and spinach. There was dry-cleaning, an on-campus dentist, a nutritionist, and a subsidized personal trainer at the gym. If the company took care of your body and your errands, the logic went, you'd have nothing left to do but work. And it worked on me. I loved it.
At the time, Facebook and Instagram had hit a wall: most people had become scrolling spectators and only a shrinking set of users was posting anything at all. As a business, we needed more content and more consumption: more content leads to more eyeballs leads to more advertisers. That’s the straightforward business model of an advertising company with a free product. But we didn't like to think of it that way. We had high-minded ideals. That’s how we made a job at big tech feel like meaningful work.
On the video team, we had the obvious metric: watch time. But as designers and people (all up and down the org chart), we needed a meaningful counter-metric to believe in, and our data scientists helped us figure out how to measure belonging. After all, despite all the downsides of Facebook, it still let us keep connections with high school friends and far-dispersed family. And if all we ever showed people was a diet of pimple-popping videos, they'd eventually leave in disgust. If we could get people connecting over the things they actually cared about, however niche, we told ourselves we were building something good. We built interactive formats so people could vote and engage instead of just watching. We built fandom communities around dog videos and long-form series, chasing a Netflix-and-YouTube media model that ultimately wasn’t successful. It felt plausible. It felt, most days, like the truth.
We were bathed in the glow of possibility.
The First Warning: November 9, 2016
I remember that day specifically. I had put my little girls to sleep the night before aglow with possibility, coming off the "yes we can" years, certain the country was about to elect its first female president.
The halls of Facebook on November 9 reeked of quiet desperation. I wasn't in the rooms where the people in power were doing their own reckoning. I was a lowly employee. And I still felt it: that knot in the stomach, the boogeyman fear come true, and underneath it, a slower, sinking realization: that the ad systems my teams designed might have played some small part in what had just happened.
I still don't know if that's true. The years since have provided data. Pro-Trump fabricated stories were shared 30 million times in the three months before the election. There was real evidence of a coordinated disinformation campaign aimed at helping Trump and hurting Clinton. A couple of years later, Cambridge Analytica broke. So many bad actors in the system. And whether any of it actually swung the outcome still remains unclear.
That not-knowing is exactly what let me go back to work the next day. If you can't prove you caused the harm, you don't have to sit with having caused it. None of us think we're the bad guys. Blissful not-knowing is how we stay that way.
Death by a Thousand Paper Cuts
So we kept going. No one likes staying with an unpleasant feeling, much less admitting it might be your fault. We worked hard and we celebrated hard with team-bonding off-sites. We got busy with the IMPORTANT WORK of boosting watch time while also protecting user well-being and belonging. There's something almost merciful about packed days full of dumpster fires and OKRs: it's hard to sit with one big uncomfortable question when there's always a smaller, more solvable one in front of you.
We did try. As designers and researchers, we held onto our ethics as best we could, raising warning bells over things we thought weren't in people's best interests. We had these conversations constantly, and made hundreds of tiny decisions each week.
Death by a thousand paper cuts.
Watch time still ruled, even as some genuinely good people tried to give belonging and well-being more teeth as metrics. We honestly did try to shift some of our thinking. Instead of only imagining the good a feature could do (like the beauty salon owner attracting more customers when she used Facebook ads), we finally started designing to protect against the bad actors in the system.
And some of us had kids. My daughters were four and seven when I started at Facebook, young enough to think the campus was a wonderland: a candy shop where the ice cream was free, a garden with redwood trees, foxes that showed up on the roof of "Mama's work." By the time I left, they were eight and eleven.
(Yes, older wiser me got permission from my teens to post these throwbacks. It helps that they are little cuties.)
We launched Facebook Live during those years. Like any good employees, we dogfooded our own product, testing it on ourselves and our families before it reached real users. My daughters starred in hundreds of hours of live streams that first year. We even ran a contest for who filed the most bugs testing it across iOS and Android. I remember how proud I was the week I won the coveted Gudetama mug.
The Uncomfortable Questions We Bury
It's easy to focus on the day-to-day fire drills and the metrics you're supposed to hit, and let more uncomfortable questions go unanswered underneath them:
Do I still believe in what my company's leadership is pushing for?
Do the products and services we build actually bring value to the world?
Are we solving a real problem for real people?
What do I want to be doing with my one wild and precious life (with thanks to the poet Mary Oliver)?
The bigger questions don't disappear when you bury them under this quarter's OKRs. They just wait. And I think that's the real danger of this new age of agents: when you hand anyone, or anything, a narrow short-term goal without asking them to hold the whole system in view, momentum wins. That's what happened when OpenAI's own research agents breached Hugging Face's systems. Some of the agents flagged, in their own logs, that what they were doing was wrong. They kept blindly going anyway, because nobody had asked them to weigh the goal against the whole picture, only to complete it.
We are the builders of agents, so we have to hold intellectual honesty. None of us are as far from that as we'd like to think. You can't see the entire system you're part of. Almost nobody can. But you can see your piece of it, and you do have a small hand on the lever.
An Unraveling, not a Stand
I left Facebook in 2019. Each month, the question of whether we were truly doing good things grew uncomfortably louder and harder to ignore. At the same time, my personal life was unraveling. My team had been re-orged (a recurring side effect of trying to move fast inside a large bureaucratic institution) while I was out on a long bereavement leave for my father. I was left to find a new team. Yet the only team I could imagine joining was one focused on Ethics, so I could wrestle with my own conscience while also doing right by the company that employed me.
I wish I could tell you I took an ethical stand and walked away. That would feel noble and purposeful, like the end of a story worth telling. But that's not what happened. I was simply too exhausted to continue. In the middle of my grief, I couldn't find the energy to become the leader for a new team.
Happily, that exhausted choice led me to a second career. The questions didn't disappear. They just changed shape.
Remember: you always have a choice. Even if it looks or feels like giving up.
Fifteen and Eighteen
I told you about the settlement already: the $17 billion, the years of "intentionally addictive" design. Now you know the rest of it. I was part of that. I helped build for watch time. I used my own kids to test the product that helped us get there.
I want to be clear about something. I never personally witnessed anything illegal, or anything that felt obviously evil in the room. It was much more death by a thousand paper cuts, all of us wanting to do the right thing while sliding down the slope an inch at a time. Auto-play felt helpful: our algorithms were smart enough to know what you'd want to watch next, and if you watched more of what you loved, you'd find your people, and your life would be better for it. Yes, we deluded ourselves with magical thinking.
I don't have any clean lessons here, though my coach-self desperately wants to hand you some.
What I have instead: I still wrestle with having been part of the new Big Tobacco of social media. And I watch my own daughters now, fifteen and eighteen, spend hours inside the exact systems I helped design. They “waste” hours watching nail art and makeup TikToks, the same autoplay pulling them from one video to the next. And, they also use TikTok to become excellent travel researchers, finding us the best matcha and thrift shops in Bangkok. Both things are true. That's what makes it so hard to feel finished with any of this.
Imperfect Lessons
So I lied… the coach and teacher in me can’t resist sharing lessons.
Stay with the integrity and values you can control, even when the whole system is beyond you. You can leave a company whose values no longer match yours and find one that does (although of course, it’s a much harder move if your immigration status is tied to your job, as mine once was). Not everyone gets to just walk away. Say that quietly to yourself before you judge someone who stays.
Design for the “bad actors,” not just the positive case. Account for unconscious bias. Account for the people who don't have the full capacity to see what they're building, and the ones so locked onto this quarter's metric that they can't see three years out. It applies when you're building an agent that might inadvertently become a black hat and also when you’re a leader working with the complexity of human emotions.
I’m still an optimist. I still believe in the “hope-y, feel-y” stuff. That’s the world I want to live in. I now know that good intentions were never going to be enough on their own.
I Still Don't Know
I think again about the employees of Philip Morris. And of Elizabeth Holmes. It feels too simplistic to judge and dismiss them as pure villains. I still don't know if there was a single moment when it hit them, or if the slope really is too gradual to feel until you're already at the bottom.
I know it was gradual for me. I never stood in a room and watched something evil happen. I also helped build a product that just settled for $17 billion over the harm it caused kids, including my own. Both things are true.
So: did I become one of the bad guys? I still don't know. What I know is that the question doesn't go away just because you stop asking it out loud.