« What is the most powerful persuasive page on your website? | Main | Small victory: Mobile Persuasion book listed on Amazon »
June 24, 2007
Persuading people to interpret scanned words
Luis van Ahn of Carnegie Mellon is a clever creator of persuasive technology. After exchanging emails with him over the last few years, we finally met in person yesterday at O'Reilly's Foo Camp. We talked about his latest project with captchas.
You've seen captchas before. They look like this . . .
Luis is now using captchas to get humans to do boring work that computers cannot do: interpreting scanned words.
In most cases computers using OCR can recognize words on a scanned page. However, once in a while, the computer can't recognize a word with confidence. That's where Luis van Ahn's captchas come in. His system takes the puzzling word and presents it to a human, who can usually recognize the word and type it into the box. This human intervention improves the quality of book scanning projects, like those championed by Brewster Kahle of Internet Archive.
We humans don't like doing boring work, such as typing in random blurry words. But once again, Luis has linked boring, pattern-recognition tasks to another goal that motivates humans. For example, people want to register for web services like Facebook or Yahoo Mail, they must complete a captcha. In this way, boring but useful work is getting done, little by little.
You'll find a longer blog post by John Murrell, which I won't rehash. Instead, I'll point out that I've written about Luis' work before, calling him the "Mary Poppins" of the Internet (remember how she got those kids to do their chores?).
I'll be watching for the next clever solution from Luis.
--BJ Fogg
Posted by BJ Fogg at June 24, 2007 06:12 AM