I came into data in 2013, which means I've now been here for something like a dozen revolutions — each one, at its peak, described in exactly those terms, each one going to change everything, most of them changing rather less than promised and a few changing more than anyone expected in directions nobody predicted. Big data. The internet of things. Digital twins. Data lakes. Data mesh. The modern data stack. Generative AI. Agents. Thirteen years of standing inside successive waves of certainty, and the single most useful thing the whole run has given me isn't expertise in any one of them. It's a feel for the shape of a hype cycle — the ability to stand inside one while it's roaring and roughly locate where I am on the curve. That feel is worth more than knowing any individual technology, because the technologies keep changing and the shape doesn't.

The shape, once you've seen it enough times

Every hype cycle I've lived through has moved through the same acts, and knowing the acts is half of not being fooled by any of them.

  • The genuine spark. It always starts with something real — a true new capability, a genuine problem newly solved. This matters: hype cycles are not cons, they're overreactions to real things, which is exactly what makes them so easy to get swept up in. Big data really did let us handle volumes we couldn't before. IoT really did connect the physical world. Generative AI really can do things that were impossible a few years ago. The spark is real, every time.
  • The universalisation. Then the real, specific thing gets inflated into a universal one. The technology that solves a problem is recast as solving every problem. This is the tell — the exact moment "this is great for X" becomes "this changes everything" — and after enough cycles you can feel it happen, like a pressure change before weather.
  • The trough. Reality arrives with an invoice. The thing turns out to be excellent for its actual use and mediocre or ruinous outside it, the early adopters hit the limits the marketing omitted, and the mood swings from "changes everything" to "overhyped failure" — which is just as wrong as the peak, in the opposite direction.
  • The quiet productivity. And then, long after the noise has moved on to the next wave, the technology settles into being genuinely useful for the specific things it was always actually good at. No headlines. Just quietly load-bearing. This is where the real value was the whole time, and it's the act nobody's paying attention to because the caravan has left for the next spark.

Where I've been fooled — in both directions

I'd be a fraud if I presented this as wisdom I've always had. I've been caught by the cycle repeatedly, and instructively, in both directions.

I've been caught by the peak — swept up enough in a wave's certainty to over-recommend something, to believe the universalisation, to let genuine excitement about a real capability tip into overstating its reach. Every honest technologist has done this, and anyone who claims they haven't is either lying or wasn't paying attention.

But — and this is the subtler failure — I've also been caught by the trough, which I think is the more damaging mistake and the less discussed one. I've been burned by an over-hyped thing, developed a reflexive cynicism about it, and under-rated it precisely as it entered its genuinely-useful phase. I called things dead that were merely past their peak, and missed real value sulking about the earlier overreach. Learning to distinguish "this was overhyped" from "this is useless" took me years, because the trough feels like clear-eyed wisdom when it's really just the mirror image of the same error — an overreaction, wearing the costume of skepticism.

What actually lets you read the cycle from inside

So how do you locate yourself on a curve you're standing inside, in real time, when everyone around you is certain? A few instruments I've come to trust:

  1. Separate the real spark from the universalisation, explicitly. For any hot technology, force the two questions apart: what can this genuinely, specifically do — and what is it being claimed to do? The gap between those two answers is the hype, measured. Naming the real capability precisely is also what inoculates you against both the peak and the cynical trough, because you're anchored to what's true rather than to the mood.
  2. Watch for the universal claim. The reliable tell that you've hit the inflation phase is the shift from specific to universal — "great for this" becoming "changes everything." When you hear it, you're near the peak, and near the peak is exactly when to get more skeptical, not less.
  3. Ask who has to run it at 2am. Hype lives in the demo; reality lives in maintenance. "Who operates this when it breaks, and what does it cost them?" is a question the peak never wants asked, which is precisely why asking it locates you on the curve so reliably.
  4. Follow the value into the quiet phase. The least glamorous instrument and the most valuable: keep paying attention to technologies after the caravan leaves, because the genuinely-useful phase — the load-bearing one — happens in the quiet, and that's where the returns actually are for anyone still looking.
A hype cycle isn't a lie you see through once. It's a shape you learn to recognise while you're standing inside it — real spark, universal overclaim, cynical trough, quiet usefulness. Miss the shape and you'll get fooled at the top and fooled again at the bottom, by the same wave.

Why the shape matters more than any technology

Here's what thirteen years and a dozen revolutions have actually left me with. The technologies were, in a sense, interchangeable — each real, each oversold, each eventually settling into its true and narrower usefulness — and if I'd bet my career on being the world expert in any single one of them, I'd have been repeatedly stranded as the wave moved on. What compounded instead, across all of them, was the meta-skill: reading the cycle, separating spark from overclaim, keeping my head near the peak and my curiosity alive into the quiet phase. That skill has never once gone obsolete, because it isn't about any technology. It's about the very human way we collectively overreact to real things, and that pattern is thirteen years old in my experience and much older than me in truth.

Which brings me back to the conviction I started with, in 2016, before I'd earned it: that data is people communicating, and that reading it well is finally a human act, not a technical one. A hype cycle is that same truth in a different costume — not a property of the technology at all, but of us, of how we talk ourselves into and out of things collectively. The person who can read that, calmly, from inside the noise, will stay useful across every wave still coming — the agents, and whatever gets called a revolution after them. The waves will keep arriving. The shape won't change. And the quiet, unfashionable, load-bearing usefulness will keep hiding, every time, exactly where the noise has stopped looking.