The Hype Cycle as a Second-Order Cybernetic System

GartnerGartner analyst Jackie Fenn published the first version of the Hype Cycle in a February 1995 research note, "When to Leap on the Hype Cycle" — a hand-drawn curve plotting a handful of technologies against a five-stage arc: Technology Trigger, Peak of Inflated Expectations, Trough of Disillusionment, Slope of Enlightenment, Plateau of Productivity¹. It became one of Gartner's most recognisable products, republished annually across dozens of technology categories. It's also usually explained as if it were simple noise around a real signal — hype overshoots actual value, then corrects. That account is too weak to explain why the correction is often so much larger than the overshoot, or why some technologies never complete the curve at all. A better account treats it as a genuine feedback system, and a specific kind of one.

Two Inputs, Not One

SeparateSeparate the curve into two distinct quantities rather than one wobbly line. Utility is the technology's actual, delivered capability — what it can really do, for whom, right now. Publicity is the visible attention it receives — coverage, investment, conference talks, the number of people who can name it. These aren't the same thing, and they don't move at the same speed. Publicity can rise from a single demo, a funding round, or a well-timed launch; utility only rises from actual engineering, deployment, and the slow accumulation of cases where the thing genuinely worked. The Technology Trigger is publicity starting to climb steeply while utility is still close to its starting point — not because early utility is worthless, but because it hasn't yet been tested against the range of situations publicity is now promising it will handle.

Why This Is Second-Order, Not First-Order

AA first-order feedback system corrects against a real, external signal — a thermostat measures actual temperature and acts on the measurement. Applied naively to the Hype Cycle, that would mean publicity is just a noisy, lagging estimate of utility, overshooting and undershooting a real quantity that stays where it is throughout. Heinz von Foerster's distinction between first- and second-order cybernetics — the cybernetics of observed systems versus the cybernetics of observing systems, where the observer's own model is part of what's being described — names what's actually missing from that account². The community isn't measuring utility directly at the Peak; it's acting on its own model of utility, built from an early, small, self-selected sample of impressive demonstrations, and that model is itself a product of the publicity that's already climbing. The model and the publicity that produced it aren't independent quantities to be compared — the model is downstream of the very publicity it's supposed to be a check on.

That reframes what happens at the Peak and in the Trough. The Peak of Inflated Expectations isn't utility being overestimated by some fixed margin — it's an overfit model, extrapolated too aggressively from too few, too favourable early cases, the same failure mode as generalising from a small unrepresentative sample anywhere else. The Trough of Disillusionment isn't utility collapsing; it's that overfit model getting falsified once broader, less curated deployment actually tests it, and the community's account of the technology's own worth gets revised downward to match. The underlying capability curve can be, and often is, still rising smoothly and undramatically through the Trough — what's crashing is the community's self-model, not the thing the self-model was supposed to be tracking. That's the strange loop that makes this second-order: the system's account of its own value shapes the investment and deployment that generates the evidence used to revise that account, rather than an outside instrument simply reading utility off a dial.

Not Every Technology Completes the Same Way

Fenn'sFenn's original curve implies one pass, ending in a plateau. Real technologies split into at least three distinct shapes once the model is corrected, and which one happens depends on what the corrected utility model actually finds:

• Recovers. The corrected model still supports real, if less dramatic, value — investment continues at a lower, steadier rate through the Trough, and the technology climbs the Slope toward a genuine Plateau.
• Recycles. A later, real capability improvement re-triggers a fresh cycle from a point the technology previously occupied, rather than the same cycle simply continuing — the system re-enters Trigger-to-Peak territory on genuinely new grounds, not a rebound from the old peak.
• Dies in the Trough. The corrected model finds the actual utility was never sufficient to sustain continued investment at all — not overhyped-then-recovering, but overhyped relative to a ceiling that was always low. Nothing pulls it back out.

Reading Seven Technologies Through Both Models takes each of these three shapes to a real example, including one technology that has done it twice.

What This Page Adds, and What It Borrows

Fenn'sFenn's Hype Cycle and von Foerster's second-order cybernetics are independent pieces of work from unrelated fields, decades apart, and neither author connects them. Reading the Hype Cycle as a second-order cybernetic system — utility and publicity as distinct inputs, the Peak as an overfit self-model rather than a fixed overestimate, the Trough as model correction rather than value collapse — is this page's own synthesis, not a claim either source makes.

Related Topics

• The Half-Life of Knowledge — a different axis on the same technologies: how long until something's superseded, rather than how far perceived value diverges from actual value along the way.
• Placing a Technology on the Half-Life Scale — a practical rubric for the half-life axis, and why foundational-tier material rarely generates a hype cycle at all.
• Reading Seven Technologies Through Both Models — the three completion shapes above, applied to real, dated cases.
• Managing the Trough: For Builders and For Learners — the practical use of the model-correction framing above.
• Funding Blue-Sky Research to Survive the Trough — the same model-correction framing applied to a government funding decision (the Lighthill Report, Japan's Fifth Generation project) rather than a market.
• Inside the Rhizome — a vendor-stakeholder is structurally a publicity-amplifying node in exactly the sense this page describes; a researcher is closer to a utility-tracking one.

References