Freebase is like a combination of Google Base and Wikipedia.
It wants to be a big structured open database. If you realize it's called Free Base, and relate that to a free & open version of Google Base, you get most of the point (assuming you are one of the .00003% of web users who have ever heard of Google Base).
But Freebase also has a presentation layer like Wikipedia, with destination pages that can be edited. It has nice tools for version control and adding semantic data to each page.
However, I think Freebase faces some key obstacles in gaining mind-share:
Freebase is an exploratory semantic web blob. I'm personally of the Clay Shirky school when it comes to semantic web stuff. I.e. it's not really good at anything, and it's a research problem to determine if it ever will be. The people that get excited about this stuff fall into a small subset of the .00003% who know what Google Base is.
In other words, Freebase has no killer-app which would compel a web user to use it. The ones it could eventually have were taken by someone else first.
With Danny Hillis as a backer of the parent company (MetaWeb), Freebase will probably end up as a good playground for the Semantic Web-o-philes out there.