Daniel Lemire's blog

, 1 min read

Coverage of the cuil search engine

It seems that the Cuil search engine is getting reactions from almost everyone. On my blog roll, about ten people have commented on it.

Here is my verdict:

  • Cuil.com claims to have outdone Google as far as recall goes: Cuil searches more pages on the Web than anyone else—three times as many as Google and ten times as many as Microsoft;
  • It is not difficult to find queries for which Cuil returns nothing, or returns far fewer pages than Google does (See what Daniel had to write about this).

They also do some semantic analysis. For example, if you type my name, then it knows that I am a famous comedian (hint: I am not!), and helps you find other famous comedians. I think they made a mistake to tie the semantics to the query, and not to particular results. You can tell me more about a given Web page, that’s useful. Guessing what my query meant? That’s reckless.


  • Cuil needs more tuning,
  • semantic analysis should not be so tied to the query terms,
  • they do not appear to outdo Google in any significant way, and certainly not with respect to recall (but this may change).

Update: Sylvie actually likes it?