Framework for a commenting service

Over at Carbon-Based Curiosities there have been some interesting discussions on a social commenting service for journal papers. In addition, PMR has been discussing the merits and best methods to do standoff annotation of PubChem.

Here's a mashup method that would be suitable for both, and reuses existing technology so that only the glue needs to be written:
(1) Create a blog called http://commentsonpapers.blogspot.com
(2) Write a greasemonkey script (or Firefox extension) that adds a link "Comment on this" beside the title or DOI of the paper of interest
(3) When you click on the link, it creates a new blog entry for that paper or else brings you to the existing blog entry.
(4) You add your comments like in a normal blog, and click Save.
(5) The script in Step 2 should also flag any papers that already have comments

Advantages:
(1) Commenting is already well developed for blogs. There is anti-spam system in place. You can subscribe to comments on your own papers, etc.
(2) Moderation of comments is possible, although the entire chemical blogosphere would have to help out on this - no moderating of comments on your own papers, of course. :-)
(3) The Greasemonkey script for displaying comments from existing blogs does many of the same steps, and has shown that it *is* possible (not just pie in the sky). The chemical blogosphere is fewer than 100 people at the moment, whereas the potential commenters on papers is much larger.

Disadvantages:
(1) Users need Firefox, Greasemonkey and the Greasemonkey script.
(2) I think that it may require a standoff site (similar to Chemical Blogspace) to gather the list of existing comments and make it accessible to the Greasemonkey script via an API. However, I could be wrong. In any case, that is no big deal.
(3) The blog username and password would be contained in the Greasemonkey script for all to see. Can varying levels of permissions be set for different users of a blog? Alternatively, the blog entries could be created automatically based on RSS feeds from the journals, and so only the comments would be possible using the Greasemonkey script.
(4) Would probably need to develop a standoff backup service for comments, in the event of people deleting them. Not a big deal - possibly has already been done.

The main thing I'm worried about though is that this has already been done by Connotea. I'm not very familiar with Connotea - is that a better route for social commenting? I know that there are already some Greasemonkey scripts for Connotea - can you comment on a paper from a journal web page, and see existing comments?