#659 1977 · Paul MacCready · Aerospace / engineering design
MacCready won the human-powered flight prize by building a plane cheap enough to crash on purpose
the problem
Every team chasing the prize kept losing months rebuilding after each crash
background
For 18 years after the Kremer Prize for human-powered flight was announced, every team that attempted it built the strongest, most aerodynamically refined aircraft its engineers could design — a reasonable instinct, since a fragile aircraft seemed like the last thing you'd want attempting a genuinely difficult, failure-prone challenge. But building for strength and optimization meant every crash, and crashes were common on a problem this hard, cost a team months of rebuilding before it could try again.
A team that could only attempt the challenge every few months, because each failed attempt destroyed months of work, could only learn from a handful of real tests across years of effort. The actual constraint on solving the problem wasn't the quality of any single attempt — it was how many attempts a team could realistically run and learn from.
what everyone would do
Build the strongest, most refined aircraft possible before attempting a flight — the instinct every team chasing the Kremer Prize had followed for 18 years, since a fragile aircraft seems like the last thing you'd want for a genuinely failure-prone challenge. It fails because building for strength means every crash costs months of rebuilding, and on a problem this hard, crashes were common — so a team could only attempt the challenge a handful of times across years, no matter how good any single attempt was.
what they saw
MacCready saw that the real bottleneck wasn't the quality of any single attempt, it was how many attempts a team could actually run and learn from — a team optimizing for survivability was optimizing the wrong variable entirely. If the aircraft was cheap and fast to rebuild instead of robust, the team could fail constantly and still make more real progress than a team protecting each attempt from failing at all.
the move
Paul MacCready designed the Gossamer Condor as a deliberately flimsy, ultralight aircraft built from foam, mylar film and thin aluminum tubing — cheap materials fast and inexpensive to repair rather than robust enough to survive hard landings. The design let his team fly, crash and rebuild within hours instead of months, running roughly 222 test flights over a period of months and refining the aircraft after each failure far faster than any team optimizing for survivability ever could.
why it works
Building the Gossamer Condor from foam, mylar film and thin aluminum tubing meant a crash cost hours to repair instead of months, so the team could fly, fail, and rebuild dozens of times in the same span a conventional team spent recovering from a single incident. Because each attempt generated real data about what worked and what didn't, and the next attempt could incorporate that lesson almost immediately, the rate of learning compounded across roughly 222 test flights rather than being throttled by rebuild time — which is why a deliberately flimsy aircraft ultimately won a prize 18 years of sturdier designs hadn't cracked.
the payoff
On 23 August 1977, pilot Bryan Allen flew the Gossamer Condor's 223rd flight, completing the required figure-eight course around two markers half a mile apart in 6 minutes and 22 seconds, winning the £50,000 Kremer Prize for the first sustained, controlled human-powered flight.
where it breaks
The approach only works when failure is genuinely cheap to recover from and doesn't carry catastrophic consequences beyond lost time — a domain where a failed attempt destroys something irreplaceable, endangers people, or can't be quickly rebuilt gets no benefit from optimizing for fast failure, since the cost of each attempt isn't actually low. It also depends on each failure producing a usable lesson that improves the next attempt; if failures are too random or uninformative to learn from, more attempts just produce more noise rather than faster convergence on a working design. And it requires the team to have a way to run attempts fast enough that iteration speed genuinely outpaces the alternative — a fast-failure approach that still takes weeks between attempts loses most of its advantage over a team building fewer, sturdier prototypes.
what came after
The Gossamer Condor's design philosophy — build for cheap, fast failure and iteration rather than for surviving any single attempt — is widely cited in engineering and design writing as a case study in reframing a hard problem from 'how do we succeed on the first serious try' to 'how many tries can we actually afford to run,' a principle since applied well beyond aviation to iterative product design and rapid prototyping generally.
references
- [1]MacCready Gossamer CondorWikipedia, 2025en.wikipedia.org
- [2]Paul B. MacCreadyNational Inventors Hall of Fame, 2022invent.org
- [3]Paul B. MacCready, Ph.D. — Father of Human-Powered FlightAcademy of Achievement, 2023achievement.org