corin_ hit the hammer on the head with that.

Sory, I should have tossed a link in when I made that comment.

There are always two words that you will see. One that is known, and the other is unknown.

Once enough users (lets use 10 for this example) type the same responce to the unknown word (and correctly identify the known word), the unknown word will move to a known status and be matched up with a new unknown word.

Thereby digitizing books! (ever wondered why one of the words is clearer than the other? Now you know!)