New Approach to the CAPTCHA Problem

There is nothing worse than receiving a few hundred automated enquiry form submissions per day generated by spammers using “bots”.   At Vertical Leap, until we put up a CAPTCHA on our enquiry forms – this is how many we were getting.

Maybe it was because we have the word “Search” in our web site title and so were getting included in the “sumbit your site to 300,000 search engines for $9.99” services, but it became unbearable.. quickly.

A CAPTCHA is an image usually with letters and numbers that is used to prevent automated use of a web site and is actually a loose acronym of “Completely Automated Public Test to tell Computers and Humans Apart” (or something similar).

Our CAPTCHA used randomly generated letters and numbers in an image file with lots of background noise.  This makes it much harder for the character recognition software used by some of these spammers.  For us – it cut out the form-spam completely (although it did increase the number of enquiries by phone we started to receive from people who couldn’t be bothered to complete the CAPTCHA box).

The problem is that the spammer’s bots are getting better and reading more complex CAPTCHA forms.  In some industries, they are replacing automated software with low cost offshore labour to complete the forms manually.  So it seems that form spam will start to rise again in the near future.

I read today that Google was working on a new twist to the CAPTCHA.  The new approach involves showing users three different images and requires the user to rotate them so that they are visually “right side up”.

Apart from photos with faces, this approach would apparently be very difficult for software to accomplish – but I think it could present users with problems too.

First, I think that CAPTCHA’s are very inconvenient.  In fact, Google’s current system is so difficult that it usually takes me several attempts to complete.  I typically get frustrated at the pure inconvenience of it and I suspect other users would as well.

Second, what are the margins of error for being “upright” – would it require tweaking left then a bit to the right and then to the left.. to make it “acceptable”?

Perhaps very large, dominant brands such as Google can get away with this sort of approach – but I suspect it will create ill-will and irritation for users of many other web sites.

Surely there is a better, simpler way to tell your human users from your computer users?

Comments are closed.