Users need to register with their real identities and are encouraged to leave short posts with references if necessary (see how to get your comment noticed below). Don't quote unpublished results or be offencive (racist, sexist, overtly commercial, etc).
Many of the comments are detailed academic evaluations of the work. Some are personal opinions and some are the lead authors summary of findings. Hopefully these comments will continue to come from the same direction; as other academics review work against alternate views, as authors update their own comments to reflect additional work and hopefully as more people say they like or otherwise the work described.
I'm not sure when the pilot phase closes and PubMed commons opens up to everyone. I can only say the sooner the better and well done to PubMed for what I believe is going to be a useful addition to literature searching.
How to join PubMed commons: Joining instructions can be found at the PubMed website, it feels a bit like the early days of gMail when only a gMail user could invite new people onto the system. I think it could grow very quickly under this model but the pilot should conclude and then access will be universal. After all not everyone is funded by NIH or Wellcome!
How to get your comment noticed: One detail I am unclear on is why the PMID is being used rather than the DOI (which I like and have posted about previously). PubMed recommend you simple use
What's on PubMed comons? Of the papers receiving most comments (early days don't forget) a couple stood out to me:
The 2011 Science paper that has already attracted a lot of negative attention, and had a response posted by the authors on Science, A bacterium that can grow by using arsenic instead of phosphorus.
A PNAS 2005 paper; Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, a very high-impact paper on anyone who's been running microarrays, RNA-seq and similar for most of the last decade.
Will PubMed commons be just another place to leave comments and will a site like Altmetric become the place we go to gather data on papers as they "scrape" information from multiple sources? I guess time will tell, for now we'll just have to keep looking and listening for comments on our research papers. You can follow PubMed on Twitter @PubMedCommons.