You go to a website. You decide to sign up for a new account there. You’re taken to a screen where you meticulously enter your details, making sure you dont leave out any required fields (or else you’ll have to retype your password… twice). And right before you are allowed to hit “Submit” you see the final challenge of registerering for something online– a box with some strange symbols all jumbled up (possibly incomprehensible upon first glance) with the instructions “type what’s in the box.”

Its not a new scenario– in fact, it’s probably something most people have had to deal with online since Captcha really got kicked off in the late 90s. In general, a Captcha is a challenge-response test that is designed to make sure the user taking the test is actually a human. This is based on the assumption that humans are better at character recognition than machines. Indeed, the algorithms for optical character recognition (OCR) wern’t very good at figuring out Captchas when they were first introduced. Therefore, Captchas originally provided a good defense against spam robots or automated programs that wished to abuse features of online services that were designed or intended to be used only by real people.

But how useful is Captcha nowadays?

With the widespread proliferation of spam and bots (especially among social networks and forums), I would say the purpose of Captcha is still very important—almost essential. For at least the near term, we will continue to need a way to verify that the people using a website are actually people. But Captcha has its problems—it’s a usability liability (people don’t want to be bothered with “guess-this-word” puzzles all out the random). Also, Captcha isn’t very friendly to the visually impaired (although some implementations offer the option of playing an audio recording of the words or letters to address this issue).

This begs the question: is there something else that can achieve this purpose better than Captcha? If so, then perhaps Captcha itself has outlived its usefulness. If not, then maybe we still need it around.

There have been lots of attempts to create alternatives, from identifying kittens to a type of image-word association.

Most attempts look OK and work OK… but not really all that much better than regular text/image Captcha. Many even suffer from the exact same problems.

And then there is the constantly improving OCR technology. There is only so much you can do to mangle an image before even humans can’t easily read it. This is a potential upper bound, and pretty soon, the algorithms might be just as good (if not better) than human recognition. Even GMails Captcha was broken recently.

So perhaps Captcha really is on its way out— it will be interesting to see what will take its place in the future. Maybe the underlying purpose behind Captcha will change somehow. Or maybe a truly innovative technology will emerge. But for now… I, for one, am glad it is around to help keep our Internets clean of the riffraff. At least for the short-term, its usefulness is certainly nothing to be contested.

This entry was posted on Friday, May 22nd, 2009 at 10:26 pm by Nick Staples and is filed under general. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.