Walter raises lots of good points (as per his usual) and I thought I would
throw out  a few comments on javascript (I can pick his brain on the rest
of them while in Toronto for the Ontario Library Association conference).
There is no doubt that javascript can be tough to integrate into web
applications, some browsers, particularly the abomination know as Internet
Explorer on OS/X, seem to be almost javascript hostile. That being said,
there are ways to selectively insert content with javascript on a web page
based on the content that is already there. You can also do some amount of
client-server communication, though this can be tricky.

One technique is to use a hidden iFrame, which is probably the way to go,
but I only recently learned about this method and I tend to use images as
conveyors of server-side information. SVG, the XML standard for images, can
be used in environments like Cocoon to dynamically create JPEGs, GIFs, or
PNGs through the magic of stylesheets. So you can, for example, feel the
results of an SQL query into a stylesheet to create a certain image based
on the rows that are returned from the database. This means you can alter
the image to indicate certain conditions, e.g. an image that is 15 pixels
wide marks that a title is in the collection, and so on. I am with Walter
that inserting links without knowing if they are useful can sometimes be a
problem, I wish, for example, that OpenURL resolvers would do more with
icons based on predetermined availability rather than the "click here and
hope" approach.

At some point, XPointer (no pun intended) will make it possible to achieve
this without using javascript at all, but we are not there yet. I do think
it is possible to construct a mechanism that will work in most browsers
that could add a "5 comments on this title"-type of link if comments exist,
and an "Amazon-like" link that reads "be the first to comment" or something
similar if no one has weighed in on the title. At least some blogs allow
notification of new postings and moderation of comments if there was a
concern that blogging spam or troublesome posters might need workarounds.
It would be nice to tap into other library review/comment repositories, and
amazon as well for something like this. One advantage we might have over
amazon is that we could use the patron database to indicate categories of
posts like "3rd-year student" without divulging identity.

Lots to think about on the train...