Designing Systems That Run on People
Luis von Ahn was born in Guatemala City in 1978 and received a computer from his mother at age eight. He came to the United States for his education, completing a PhD in computer science at Carnegie Mellon University, where he joined the faculty and was named a MacArthur Fellow in 2006.
His first widely deployed idea, developed around 2000, was the CAPTCHA — a reverse Turing test. Where Alan Turing proposed a test in which a human judges whether a hidden respondent is a machine, a CAPTCHA inverts this: a machine judges whether an anonymous user is human, by posing a problem that exploits the residual gap between human perception and machine perception.
Von Ahn then reframed that gate check as wasted labor. At roughly 200 million solves per day and about ten seconds each, CAPTCHAs consumed on the order of half a million human-hours daily. reCAPTCHA captured that surplus: it paired a control word, known to the system, with an unknown word on which optical character recognition had failed during book and newspaper digitization. The control word authenticated the user; the unknown word was transcribed as a side effect, with agreement across multiple independent users serving as the accuracy check. Google acquired reCAPTCHA in 2009 and applied it to the Google Books and newspaper archives.
He generalized the principle as human computation: architecting systems in which incidental human effort, aggregated across a population, performs work that remains beyond software. In 2011 he co-founded Duolingo with Severin Hacker on a related premise — that the ability to pay should not determine access to language education — building a platform that is free at the point of use and now serves hundreds of millions of learners.
