What I read for fun (and steal for work)

People assume the sci-fi and fantasy shelf is where I go to stop thinking about work. Mostly it’s the opposite. Good speculative fiction is obsessed with the same questions I deal with running operations and building GTM systems: how do you design a system you don’t fully trust, how do you make decisions with incomplete information, how do you build something that survives contact with people who don’t behave the way the spec says they will.

A few examples that keep showing up in how I actually think about work:

Hyperion is structurally about six people each telling their version of the same story before you learn what’s actually going on - and buried in that story is the TechnoCore, an AI civilization that’s been quietly steering humanity’s decisions for centuries without humanity noticing the hand on the wheel. I read that book years before “AI alignment” was a phrase anyone outside a research lab used, and it’s aged into the single most relevant thing on my shelf to the AI race happening right now: the risk was never a robot uprising, it was systems optimizing for goals we didn’t fully understand we’d handed them.

The Will of the Many is a book about hierarchy as a literal, mechanical system - power flows upward through a chain of people who’ve each ceded a portion of their own will to the level above them. Strip the magic out and it’s an org chart. What I actually took from it is the protagonist’s discipline: he’s operating inside a structure he didn’t choose and can’t openly fight, so he wins by understanding the system better than the people who built it. That’s most of operations work. You rarely get to redesign the org from scratch. You get good at finding leverage inside the one you inherited.

Mistborn gets recommended as “a magic system with actual rules,” and that’s exactly why it’s useful outside the genre. Sanderson’s rule is that magic has to follow internally consistent laws or the story stops meaning anything - and that’s the same discipline that makes a scoring model, an automation workflow, or a prompt chain trustworthy instead of a black box. If the rules aren’t consistent, nobody can reason about the system, and nobody will trust it enough to rely on it.

The Caves of Steel is Asimov using a murder mystery to ask how humans and machines build enough trust to actually work together, at a moment when both sides have good reasons not to. That’s the entire current conversation about AI adoption inside real companies, minus the trench coat.

I don’t reread any of these looking for work lessons. I reread them because they’re great books. But the reason they stick with me long after I’ve closed them is that the best speculative fiction isn’t really about the future - it’s about systems, incentives, and trust, wearing a future as a costume. Those are the same three things I spend my working life on. It would be strange if reading about them for fun taught me nothing.