Brian Reed: Eliza, hey — I had a weird moment this morning, I was trying to book a conference room through my company's system and I needed three approvals for a one-hour slot, and I just — I sat there thinking, someone designed this.
Eliza Ward: Someone absolutely designed it. And they probably thought it was an improvement over the chaos before it.
Brian Reed: Which is — hang on, that's actually the thing, right? Because that's what I want us to dig into. Is the slowness baked in, or did someone just keep adding layers without noticing?
Eliza Ward: Both, kind of — wait, no, let me be more precise. The layers accumulate because they each solve a real problem. A reporting hierarchy, an approval process, a cross-team dependency check — each one is a coordination mechanism, and each one trades speed for control. That tradeoff is structural. Max Weber identified this a century ago: bureaucracy is genuinely efficient at coordinating large groups, and it is genuinely rigidity-producing. Those two things are inseparable.
Brian Reed: So this isn't new thinking at all.
Eliza Ward: Not remotely. Burns and Stalker put a name to it — mechanistic versus organic organizations. Mechanistic: hierarchical, slow, stable. Organic: flat, fast, dynamic. And their finding was that no large organization stays organic. You scale, you mechanize, that's the structural pressure.
Brian Reed: And that mechanistic drift — so where does the math actually enter it? Because you're saying structural pressure, but pressure feels like a metaphor. Is there a number behind it?
Eliza Ward: There is, and it's — okay, this is the part that stops being philosophy. Coordination costs are estimated to scale at roughly headcount to the power of 1.5 to 1.8. Revenue per employee grows closer to headcount to the 0.9. So the gap widens by arithmetic, not by anyone making a bad call. Picture a compliance officer at a 4,000-person fintech — Thursday afternoon, she spots that a single API change touches payments, data governance, and legal. She doesn't make that call herself. She schedules a meeting, that meeting surfaces three dependencies nobody had mapped, those dependencies each need their own sign-off chain. Decision velocity — the actual time from identifying the problem to executing anything — just collapsed. And nobody did anything wrong.
Brian Reed: Wait — where do those exponents come from? Like, 1.5 to 1.8 sounds precise. Is that a controlled study or is someone's pattern-matching dressed up as a law of physics?
Eliza Ward: That's — yeah, I have to be honest there. The sourcing is fuzzy. It's empirical pattern-matching across organizations, not a controlled experiment. The combinatorial logic is solid — as headcount grows, potential communication pathways multiply combinatorially, so the coordination load can't be linear. That part holds. The specific exponent? Directionally right, precision uncertain.
Brian Reed: So if the exponent is uncertain — is some organization actually escaping the curve, or do we just not have clean enough data to see whether they are?
Eliza Ward: That's the crack, and I don't want to paper over it. What we can say is that adding hierarchy, standardization, cross-team dependencies — organizations add those not to get better, just to hold function at scale. That's coordination overhead as a scaling law. Whether anyone's genuinely bending the curve versus choosing deliberately where the friction lands — actually, that's where the Spotify Model becomes really important, and what Anders Ivarsson said later about it is going to change how that whole question looks.
Brian Reed: The part I don't get yet is whether agile-at-scale frameworks are actually reducing coordination overhead or just relocating it.
Eliza Ward: Relocating it — that's exactly the right word, and the Spotify Model is the proof. Squads of six to twelve people, acting like mini-startups, grouped into tribes, chapters for professional development, guilds for knowledge sharing. Clean architecture. Anders Ivarsson co-authored the documentation. And then later — publicly — he acknowledged it was more aspirational than a true picture of how Spotify actually operated, especially once cross-team dependencies started multiplying.
Brian Reed: Wait — the co-author said that? Not a critic, the person who built the thing?
Eliza Ward: The co-author. Which means — okay, that's not a footnote. That's the most-cited counterexample to the inevitability argument, and its own architect walked it back.
Brian Reed: So squads are mini-startups until they need to talk to each other — and then you've just rebuilt the org chart with cooler names.
Eliza Ward: That's — yeah, that's basically it. The overhead relocates, it doesn't disappear. SAFe, LeSS, the Spotify Model — they're genuine partial solutions, they shift where you sit on the speed-control spectrum. But cross-team dependencies regenerate the coordination drag the framework was designed to replace. You can't structural-design your way out because the dependencies are load-bearing.
Brian Reed: So Ivarsson's admission isn't evidence that the design failed — it's evidence the tension is real even when the design is good.
Eliza Ward: That's the hinge. The Spotify Model doesn't disprove the structural tension — it demonstrates it. The most carefully engineered escape attempt hit the same wall. That's not a design flaw you iterate away. That's the thing itself.
Brian Reed: Fine — but here's what I still don't know. How many leadership teams are actually making that choice consciously versus just waking up one day and realizing the org chart made the decision for them? Like, is there any evidence that the companies moving faster at scale chose to, or did they just get lucky with what they happened to build first?
Eliza Ward: I don't have a clean answer to that. The deliberate design case is real — you can meaningfully shift where the friction lands even when the math is against you. But whether anyone sat down and said, we're choosing this consciously — I mean, versus it just crystallizing around early structural defaults — I genuinely don't know how you'd measure that from the outside.
Brian Reed: That might be the most honest place this lands. The tension's real, the severity isn't fixed — and we mostly can't tell who chose it.
