Cameron Neylon comments interestingly in Nature on the intellectual property drawbacks of publications that are free to access but not to reuse: "Science publishing: Open access must enable open use".
The success of PubMed Central and of other disciplinary and institutional repositories illustrates a weakness. Although millions of articles are accessible to read, the majority of them cannot be used for anything except reading. If, for instance, you wish to index all the gene names in a set of papers, put them on a website, translate them, use text or images in a summary or even just print out several copies of the collected papers, you are limited to a much smaller set of around 500,000 articles that carry a Creative Commons licence (see go.nature.com/heaqoe). For any commercial purpose, which could include simply making copies for a class or company meeting, one is restricted to the smaller subset of papers that have a CC BY licence.
A heated discussion arises in the comments section. I would point out that making an index of gene names does not violate copyright in the U.S., and many other data reuses are perfectly consistent with publisher copyright. Image reuses are more important to me, as the restrictive copyright terms on distributing images of fossils on a website have severely restricted what I have been able to do here.
I do not presently have much under a Creative Commons license, but am exploring this option for some upcoming projects.