cuil is definitely adding something to the web search experience, especially if you want to find something you were not looking for.
You may be "Search(ing) 121,617,892,992 web pages", but if you take the smaller roads, you are likely to find many parasites* on the top of the list, and to turn many beautiful but mostly useless** pages before bumping into something familiar.
Don't try zooming : too many words in the search box, and you're told "We didn’t find any results". "Too many" can start with two. Even one. Actually, the "search" feature can be quite frustrating if you expect to do some googling***.
To me, cuil is not really competing directly with Google. It's more about contexts than words, more a media than a search engine, like opening a newspaper in Borges's library of Babel (without the full multilingual rainbow). Beyond randomness, there is definitely some form of intelligence, and a rather simulating one, but the tool needs some more practice, some more neuronal connexions.
So let's help this newborn grow up and feed it with more of our searches.
* definitely a revenge for fake websites traditionnally barred by other crawlers : empty shells piggybacking on the fame of others to syphon their visitors
** except for announcers, maybe. OK for limiting the number of answers on one page, but one or two items per pages, often the same from one page to the next one...
*** the most comical bug : you type an URL and it doesn't appear in the results