There’s a cool story in the New York Times Science section about a reverse Turing Test. Instead of the classic Turing test — a contest where the computer tries to dissemble as a human — it’s a test that poses the opposite condition: The humans try to prove that they’re actually, well, human.

The story begins with Yahoo. A few years back, they were having problems with spambots. Spambots were logging onto free Yahoo mail accounts and sending out piles of crap — often hundreds of times a second. Yahoo seemed powerless to stop them; how do you prevent a robot from signing up for an account?

By imposing a test that separates humans from robots — definitively. After all, there are certain skills humans have that computers have never been able to emulate:

… in many simple tasks, a typical 5-year-old can outperform the most powerful computers.

Indeed, the abilities that require much of what is usually described as intelligence, like medical diagnosis or playing chess, have proved far easier for computers than seemingly simpler abilities: those requiring vision, hearing, language or motor control.

“Abilities like vision are the result of billions of years of evolution and difficult for us to understand by introspection, whereas abilities like multiplying two numbers are things we were explicitly taught and can readily express in a computer program,” said Dr. Jitendra Malik, a professor specializing in computer vision at the University of California at Berkeley.

Precisely. Computers can multiply million-digit numbers in an instant, but do an incredibly crappy job at recognizing shapes like bees or cars or words. Humans are precisely the opposite. So Manuel Blum at Carnegie Mellon University developed a test called the Completely Automated Public Turing Test to Tell Computers and Humans Apart, or CAPTCHA. A student of his wrote — Luis von Ahn — wrote a progam that takes a random word out of the dictionary, then streches and skews it a bit. Computers can’t recognize the skewed shape; but humans can easily read the word. Yahoo now uses this test for all its free email: If you can figure out their test words, you’re a human and get an account. Otherwise, you’re a ‘bot, and they lock you out.

It’s incredibly cool. But it’s also already being hacked — as I wrote in Wired magazine two months ago. It seems that spammers have figured out an end-run around the Yahoo system:

Here’s the weird thing: Some purveyors of porn developed a way to fight back. They rewrote the spambot code so that when the bots reach the visual recognition test, a human steps in to help out. The bots route the picture to a person who’s agreed to sit at a computer and identify these images. Often, insiders say, it’s a hormonal teen who’s doing it in exchange for free porn. The kid identifies the picture, the spambot takes the answer, and – bingo – it’s able to log in. “It’s the only way we know for getting around the picture test,” says Luis von Ahn.

And the kicker:

Now consider how deeply strange this is. Instead of a machine augmenting human ability, it’s a human augmenting machine ability. In a system like this, humans are valuable for the specific bit of processing power we provide: visual recognition. We are acting as a kind of coprocessor in much the same way a graphics chip works with a main Pentium processor – it’s a manservant lurking in the background, rendering the pretty pictures onscreen so the Pentium can attend to more pressing tasks.

… Techies from Nikola Tesla to Bill Gates are famous for cheerily prophesying the day when computers will do all the drudge work, leaving us humans to dedicate our magnificently supple brains to “creative” tasks.

Except, as our machines get smarter and faster, our much-vaunted creative powers may not be so valuable. What’s more useful, in man-machine systems, is our flexibility – our ability to deal with periodically messy, wrenching situations. We won’t be doing the brain work; we’ll be doing the scut work.