There’s a talent war happening in AI right now, and it’s changed the tone of a lot of career conversations.
I’ve been thinking more explicitly about the framework I use for deciding where to work and when to stay put. I realized later on a walk home from the gym that I’d never actually written it down. This is that post.
Three dimensions:
- Meaningful relationships
- Meaningful work
- Mark to market
I didn’t invent the first two. I stole them from Ray Dalio’s Principles back in 2017, when I was carrying the hardcover on the subway every day like a brick-sized personality test. I couldn’t put it down. Those two ideas stuck. The third, mark to market, I treat as its own axis, even though some people fold it into “meaningful work.” I think that’s a mistake, and it’s the one most people get wrong.
Meaningful relationships
Who are you working with? Do you like them? Are you energized by their ideas, challenged in the right amounts, curious about what they’re building? I’ve had the privilege of working closely with old friends, and I’ve worked on teams where I barely knew anyone. I do fine in both. But it’s night and day. Working with people you’ve known for a decade, there’s nothing like it. The longer I work with certain people, the more I want to work with them. It’s a sunk cost, but in a virtuous direction.
I keep tabs on people I’ve worked with before and would want to work with again. Following what they’re building, grabbing coffees to stay in touch. Some of the companies I’d consider joining are literally only because those people are there. I don’t care about the business at all. The question is whether they can meet the bar on the other two dimensions. Relationships are a force of gravity. They pull you in directions that have nothing to do with the rational calculus you’re supposed to be running.
Meaningful work
I’m an empiricist. I like to iterate. The only way to iterate without driving yourself insane is to have a feedback mechanism grounded in truth, ideally a fast one. I care about this so much that my last startup was literally a tool for it: an AI copilot for analyzing user interviews. We launched in under six months and hit #2 of the week on Product Hunt. The product didn’t survive, but the obsession did. I want tight feedback loops with real users giving real reactions. Everything else (sales numbers, market penetration, board decks) is downstream.
Sean Ellis has this test where you ask users, “How disappointed would you be if you could no longer use this product?” If 40% or more say “very disappointed,” you have something. I want to know exactly who’s going to be disappointed if the work I’m delivering gets thrown in the garbage. That’s meaningful work to me.
Mark to market
Mark to market means repricing an asset to its current market value. Not what you paid for it, not what you hope it’s worth. Applied to your career: periodically finding out what you’re actually worth on the open market. Most developers skip this entirely. Especially the ones who love their work and like their teammates. Game developers are the clearest example: an entire class of people who have been systematically exploited because they overvalue meaningful work and meaningful relationships. They accept below-market compensation because they’re passionate, and the industry has learned to price that passion into the deal.
I think mark to market is a mental health imperative. I think about it in terms of responsibility. If I’m materially below market and I don’t know it, I’m not just subsidizing my employer, I’m pulling money away from my family, my savings, and my margin for risk. In a startup, especially, that hidden subsidy can get large without anyone ever explicitly naming it. And there are only so many days you can live with that kind of cognitive dissonance before you burn out or worse.
How does that happen? A few ways. Sometimes you believe in the business and it checks the boxes on relationships and work, but you join without a real outside reference point because you don’t have time to generate competing offers. Sometimes you can’t properly value the equity on offer because at an early stage you’re working from rough benchmarks and trust. And sometimes the market moves dramatically after you join. That’s how you end up off market without being stupid about it.
Mark to market is a process, not a number. One offer is noise. Two is a better signal. Three is even better. The goal isn’t to leave. It’s to have real information. The last time I ran this process seriously, I got much clearer information than I had going in, including evidence that the market was paying materially more for comparable work than I had assumed. Going through it was exhausting. You’re running a sales process on top of your actual job, but the output was clarity I couldn’t have gotten any other way. I also screwed part of it up, and the mistake taught me something about the hardest part of mark to market: how much to disclose.
The obvious pushback: this is mercenary. Loyalty matters. You’re burning bridges by constantly testing the market. I think loyalty is real between people. I think that’s one of the best parts of working on anything hard with other humans. But loyalty to a corporation? Corporations are machines. They optimize. I know this because I’ve been on both sides. I ran two rounds of layoffs at a previous startup. It’s brutal, and it’s not personal. That’s the whole point. And it’s not just a survival move. Look at the big tech layoffs of the last few years: record quarterly earnings and thousands of people cut in the same quarter. Companies optimize. That’s what they do.
My first experience on the receiving end was at 18, doing level-two tech support for the Compaq Presario. Compaq merged with Hewlett-Packard and I was one of a dozen people cut from the call center. At the time I thought it was my performance. It wasn’t. I was a line in a spreadsheet.
Would they be loyal to you in the same way you’re loyal to them? The answer is obviously no, and I say that as someone who has been the “they.” Being loyal to a machine that may one day optimize you away is irrational. Equity is often used to soften that reality. “You’re not underpaid, you have upside.” Sometimes that’s true. But the question isn’t what the equity might be worth in the best case. The question is what it’s worth on a risk-adjusted basis across plausible outcomes. How likely is a meaningful exit? How much dilution is ahead of you? What does your stake look like in the median case, not the fantasy case? If you’re taking lower cash comp in exchange for upside, do the math yourself with clear eyes.
Mark to market doesn’t require disclosure. Whether to tell your manager is a separate tactical question, and the right answer depends on trust, psychological safety, and how much downside you can absorb. For many people, the quiet version is the right version: run the process, get the information, and decide what to share afterward.
The framework doesn’t give me an answer. It gives me inputs. Meaningful relationships matter. Meaningful work matters. But if you’re a developer and you haven’t actually gone through the process of getting multiple signed offers recently, you probably don’t know what you’re worth. That’s not a moral failing. It’s an information gap, and it means you’re making a major economic decision with incomplete information. That gap costs real money every pay period, compounding.
Nobody’s going to make you do this. That’s the problem. There’s no deadline, no penalty for skipping it, no form in the mail. The cost is invisible: it’s the raise you didn’t know you were owed, compounding every pay period. Mark to market is the only way to see it.
Process note: Drafted from a voice memo. Used AI for transcription and editing. All claims are mine; mistakes are mine.