Why is it of interest that someone now writes this on his blog? From the start of the Netflix prize, there were countless assertions and refutations that movie information would be helpful for better predictions. You can read all about it in the Netflix prize forum.
http://www.netflixprize.com/community/
This post makes the good point that companies would have far better return if they look for ways to collect better input data rather than poring all the effort into getting minuscule improvements from the generic recommendation algorithms.
I look at how IMDb treats their data and laugh. It's basically an excuse for an API of 1990. This makes their system look like a big wooden network of pulleys and ropes. And I thought it was an Amazon company...
It's a decent API wrapper written in Python. I've messed around with it some. It's pretty cool. Although the licensing terms from the IMDB's webite were pretty strict, I thought. Basically, if you used their data for anything resembling profit, (i.e. the Netflix prize) you've violated the TOS.
