Maya Chen: Can I just sit with something strange for a second. Everett Rogers — 1963, 'Diffusion of Innovations' — he maps out this whole framework, innovators at two-and-a-half percent of the population, then early adopters, majority, laggards, and this clean S-curve connecting them. And honestly it's kind of beautiful as a shape.
Dr. Nathan Hayes: Descriptively, yes. It holds.
Maya Chen: But then — Tesla, seventeen years to its first million cars. iPhone, four years. Both S-curves. And I keep thinking, what does that mean for anyone trying to actually use this framework? Like, where do you put the four-to-seventeen-year gap in the model?
Dr. Nathan Hayes: You don't. That's — actually, no, here's what's strange about it. Christensen, Perez, Wardley — working from completely different methods, different eras — they all independently recover the same S-shape. The pattern is robust. But adoption timelines vary by factors of ten across technologies. The curve is real; the clock is ungoverned by it.
Maya Chen: So it's a reliable map of the wrong thing.
Dr. Nathan Hayes: It maps the shape of every technology revolution. Not when yours will happen. And that distinction — that's actually where the forecasting failures live.
Dr. Nathan Hayes: Now, the forecasting failure is — it's actually mechanistic. The early phase has this structural friction baked in. High costs, immature infrastructure, no standards yet, almost no skilled labor who knows how to use the thing. You're stuck with maybe two-and-a-half percent of the population, the innovators, and everything is fighting adoption at once.
Maya Chen: Which looks, from the outside, like the technology is failing.
Dr. Nathan Hayes: Exactly. And linear forecasting reads that slow early slope and just — extends it forward. Constant growth rate assumed. Which means you systematically underestimate what's coming when the feedback loops kick in.
Maya Chen: Wait — what actually kicks them in, though? Like, is it one thing or is it several things compounding?
Dr. Nathan Hayes: Both, and that's the — actually, no, let me be more precise. You get learning-by-doing first. Cumulative manufacturing experience drives costs down, so the technology gets cheaper, which pulls in more buyers, which drives more production, which drops costs further. And then layered on top of that you get network effects — the thing becomes more valuable because more people have it. Telephony is the canonical case. A telephone is worthless if you're the only person with one. Electrification took decades in the early phase partly because the distribution infrastructure had to be built alongside adoption, not before it.
Maya Chen: Hmm. So the infrastructure and the adoption are sort of... dragging each other forward at the same time.
Dr. Nathan Hayes: Right — and linear models treat those as separate. That's the seduction. The S-curve is more accurate than linear forecasting, but that's — I mean, that's a low bar. Descriptive accuracy creates false predictive confidence. You can name the mechanism precisely and still not know when it fires.
Maya Chen: Okay but I want to make this land for a second — not abstractly. There's a woman I keep thinking about. Early fifties. Her kids had smartphones, her husband had one. She held out until 2015. Seven years after the mainstream tipping point. Not because she couldn't afford it, not because she didn't understand it. She just... didn't want it. And then — her job started requiring email-on-the-go. Her bank phased out the old login system. Her granddaughter couldn't send photos directly to anything else. One day she wasn't choosing anymore.
Dr. Nathan Hayes: The network removed the exit.
Maya Chen: Right. And that's — that's switching costs, but in reverse somehow? Like, the cost wasn't entry. The cost became staying out.
Dr. Nathan Hayes: That's actually the underappreciated symmetry in switching costs. The same force slows early adoption — the economic, social, cognitive cost of leaving what already works — and then later it stabilizes the plateau. Both directions. Resistance to entry, then resistance to exit. Now, what's interesting about smartphone adoption specifically is that the iPhone's early acceleration wasn't purely network effects. The prior technology, the flip phone, was already mature. Users were replacing phones every eighteen months anyway. Apple's switching cost for the phone function itself was near zero.
Maya Chen: Wait — so the incumbency barely existed?
Dr. Nathan Hayes: For that one dimension, yes. Tesla's situation was structurally opposite — buyers had to abandon a decades-old refueling infrastructure entirely. Every gas station was a switching cost. That's not — I mean, that's not just habit, it's embedded physical reality.
Maya Chen: And those same lock-in forces — at the plateau — that's what Speelman and Numata are pointing at in 2022, right? Their five-phase model, the 'Theory of Rapid Transition' — the whole point is that patience alone doesn't move you through phases. You have to actually target the specific friction.
Dr. Nathan Hayes: Yes, and — what I keep turning over is that even granting Speelman and Numata, even granting that Rogers described this in 1963, that Christensen corroborated it from competitive strategy, Perez from economic history, Wardley from mapping — four completely independent methods, same shape — we still cannot tell you when. The S-curve is a reliable description of how technologies spread. It is a dangerously unreliable tool for predicting when they will. And that gap — that specific gap — is where the real strategic failures actually live. Not in failing to believe the curve exists. In believing it tells you more than it does.
Maya Chen: Mm. And the linear forecasting doesn't die because organizations don't know better — I mean, they do know better, some of them — it's that confident linear projections are what gets rewarded. Structurally. Politically. The S-curve answer is 'it depends on variables we can't yet measure.' That doesn't fit in a board deck.
Dr. Nathan Hayes: Correct. So the honest question — and I don't have a clean answer — is: if the curve can't tell you when, what signals actually should you be watching? Cost trajectory in real time. Network density at the core action level. Infrastructure maturity. Those are — I mean, they're measurable in principle. But knowing which one is the binding constraint, for your technology, right now — that's what the framework doesn't hand you.
