Amara's Law
We overestimate what technology will do in 2 years and underestimate what it will do in 10.
Why PMs should care
This is the single most useful law for roadmap planning, and the one most consistently ignored — by both the true believers and the sceptics in any given technology debate.
The exciting new tech your team wants to bet the company on for 2027 is probably going to be a disappointing footnote by 2028 — and quietly everywhere, taken for granted, by 2035.
The right move is to resist both the hype ('we need an AI strategy by next quarter') and the dismissal ('this is just a fad, let's ignore it'). For most products, the answer is to hedge: run small, cheap experiments that keep you learning while the short-term hype plays out, so that you have real capability and real judgement when the technology actually becomes important five to ten years later.
Amara's Law is also why you should never extrapolate a two-year trend into a ten-year strategy. The line at year two is mostly hype, and the real ten-year shape looks nothing like a straight line.
Example in product work
Chatbots, 2015–2025. In 2015, chatbots were going to replace every app: Facebook M, Google Allo, the 'conversational UI revolution', every product manager in the Valley writing a Medium post about 'bot strategy'. By 2017, the chatbots were a joke — Siri still couldn't set a timer reliably, Allo had been discontinued, M was quietly shut down. By 2025, conversational interfaces were a load-bearing part of most products, just via large language models instead of rule-based bots. A PM who bet hard on chatbots in 2015 shipped waste. A PM who dismissed them in 2017 as 'that thing that didn't work' missed the real wave eight years later. The right position in both years was the same: small experiments, retained learnings, no committed roadmap.
Google Glass → Meta Ray-Bans. Google Glass is Amara's Law in headset form. The 2013 launch was hyped as the next computing paradigm — and it was so dramatically not ready (social awkwardness, 'glasshole' backlash, privacy panic, two-hour battery life) that the product was mothballed in 2015 and became shorthand for premature tech. By 2017 it was a meme. But by 2024, Meta's Ray-Ban smart glasses had quietly shipped a million units and crossed into normal-person territory, and Apple's Vision Pro had defined an entirely new category above it. The 2013 hype was too early; the 2017 dismissal was too early in the opposite direction. A team that had held a small, patient position on the category the whole time — experimenting, learning, not committing — was the one ready to move when the real window opened.
What to do when you see it
- The exciting tech your team wants to bet the company on for 2027 is probably a disappointment by 2028 — and quietly everywhere by 2035.
- Resist both the hype ('we need an AI strategy by next quarter') and the dismissal ('this is a fad').
- The right play for most products is to hedge: small, cheap experiments that keep you learning while the short-term hype cycle plays out.
- Never extrapolate a 2-year trend into a 10-year strategy. The 2-year line is mostly hype, and the real 10-year shape is nothing like a straight line.
Sources & further reading
- Amara's Law — Matt Ridley's blogIncludes Paul Saffo's firsthand account of how Roy Amara actually articulated the law — Saffo was Amara's direct colleague for over a decade at the Institute for the Future.
- Gartner Hype Cycle — GartnerGartner's official framework, which operationalises Amara's Law into trough-of-disillusionment and plateau-of-productivity stages.