Not really, Big-O has its places. I think his point in that area was that the constant factors do matter, and are glossed over (by definition) in Big-O. Reminds me of DJB's book on high-speed crypto, where he talks about this issue: http://cr.yp.to/highspeed.html