You know, they look like this:
Facebook actually does an excellent job; it's CAPTCHA uses two English words, "contribute" and "of," which a real human being should be able to read. Psychologists and computer scientists alike agree that human beings have an incredible ability to recognize objects even when skewed.
In other words, we can read the letter "f" in a million different fonts, even though no two "f"s look the same. And let's face it, cursive "f"s look nothing like print ones. But, having learned just a few basic examples of "f," we can extrapolate what is and isn't one. That's the basis of these tests: we can extrapolate, computers cant. It really throws 'em off to have wavy letters with lines thru 'em; they can't figure out the letter that's buried in there.
The point, ultimately, is to stop people from creating programs to automatically post things to websites; the CAPTCHA ensures that a human being is doing it manually. This reduces spam.
So, what's the problem? Well, while Facebook may have a decent CAPTCHA system, not everyone does. In particular, technological superstar "Google" has about the worst verification system invented. Could you solve this?
Right. Didn't think so. And that makes the internet frustrating. So what's the solution? I say, a new CAPTCHA system. Many have been proposed, but few implemented. All we need to do is identify something computers can't do that humans can. Letter recognition was a nice idea, but it's a little confusing. What else can we use?
Well, we know that people can recognize things that computers can only, at the very best, compare to a database of existing knowledge. The nice thing about letter CAPTCHAs is that there is no database of skewed words; computers have nothing to base their answers on.
This means that some proposed ideas, such as asking users to identify famous paintings, isn't going to work. Computers can match the CAPTCHA to a database of existing images and look up the most relevant related text in order to determine its title. Besides, people aren't smart enough to handle this system in the first place.
First off, what in the world is that second image? And secondly...this is madness! That's a really inefficient system for clearing a security checkpoint - it takes too long, it's too confusing, and no one above a certain age is going to have any idea what's going on.
My solution: (it's mine, mine I say; don't even think about stealing it)...Cartoon Verification.
Oh yes. It's brilliant. You see, computers may be able to compare real-world images to a database, but what about artistic (original cartoon) representations? A human being can extrapolate the features in a caricature; a computer can't identify them. And it's so easy for people. Check it out:
|Verification Technique: "Please enter the number corresponding to the penguin."|
The beauty of the system? Any 5-year-old can get by, and no bots can. It's super-easy for people. And, most importantly, it's nothing like this:
Google, you're ridiculous. This post, brilliant. Now quick, someone find me a patent lawyer.