Blog commenters have long been familiar with CAPTCHAs, small automated tests to prove the humanity of an activity. Blogs usually require them to prove that the commenter is not a posting program, as such programs usually exist purely to post spam.
Most existing ones involve reading a wavy, distorted word. There are OCR programs that could read straight, clear text, but the wavy and distorted letters are currently only comprehensible to humans. Blind users are forced to prove their humanity through other tests, such as sound-based comprehension tests. The same robot-fooling aspect of the test foils the unfortunate person's screen reader.
One group of people wrote a new CAPTCHA, Resisty, which requires users to identify the color bands on a resistor. Its value is not its CAPTCHA value, which is low because it requires not only vision, but color vision, to pass. (Color blind or completely blind people would require a different test.) Its value is that if it is cracked, as spambot writers want to do, they now have a program that can automatically identify and classify resistors. Such a program would further the automation of making electronic devices. Whereas if the conventional wavy-text program is cracked...you have a slightly slightly better OCR. "Oh joy."