I'm looking for inspiration here. I need to employ some sort of human verification for my website, but the most common method these days (asking users to type the letters & numbers they see in an image into a text input box) seems a little rubbish - I find it hard sometimes to work out what the letters & numbers are.
There must be a better way!
I've had a few ideas, the best one seems to be to show users a series of images (4-6), and ask them to answer a question based on the contents of the images, such as:
(show some geometric shapes) "Which image has 3 sides?"
or
(show picture of animals) "which animal can fly?"
This has the advantage of being easy to program, and hopefully easy to pass.
Can anyone think of any other approaches to this problem? Or possibly spot flaws in the system outlined above? Is it possible to make such systems both easier for humans to pass, and harder for bots to pass?