Slashdot Mirror
Human vs Computer Intelligence
DrLudicrous writes "The NYTimes is running an article regarding tests devised to differentiate from human and computer intelligence. One example are captchas, which can consists of a picture of words, angled and superimposed. A human will be able to read past the superposition, while a computer will not, and thus fails the test. It also goes a bit into some of Turing's predictions of what computers would be like by the year 2000."
GetKarma () {
The NYT Article through Google
}
Which implies that the deaf are dumb.
Da-dump ding!
"Times have not become more violent. They have just become more televised."
-Marilyn Manson
That makes the phrase "deaf and dumb" a little redundant, then, doesn't it?
All I want is a kind word, a warm bed and unlimited power.
Which is of course why I added the silly little drum-drum-cymbal at the end of the post... implying that it was a dumb joke on my part.
"Times have not become more violent. They have just become more televised."
-Marilyn Manson
A couple of years ago I dreamed up (independently, I think) a captcha-like encryption algorithrm that I suspect would be difficult for computers to crack.
It works like this. You start with "plaintext" in the form of an image (e.g. JPEG) of the message in some human language.
Now you "perturb" the message, meaning you visually distort the image, so that the letters and words are very difficult for a computer to recognize. (This is the captcha-like part). You could put noise in the image, or make it look like each of the words was cut from a magazine ad, or perform topological distortions, or a mixture of these approaches, etc.
The result is easy for a human to read but very difficult for a computer.
Finally you "shred" the image. Just like a shredder for paper documents, you "shred" it (digitally) into some large number of image strips, each of the same width. The width of a strip might be a few pixels to a few 10s of pixels depending on the resolution of the original image.
The encryption key is the reordering rule that pastes the strips or pieces back together in the correct order to reconstruct the original image.
It seems to me that this algorithm would be extremely difficult to attack by brute force.
Why? Because I don't see how you would program a computer to test a proposed reordering for "correctness". How would you code a statistical or other test that was supposed to determine that some ordering of the strips was "better" than some other ordering? How would the computer know?
Of course the information density is not very high compared to conventional schemes.
But it's cute. Am I missing anything?