RSS is pretty good for news sites. I've been making use of RSS feeds for some time to unify a bunch of different news sources onto one site on my LAN, without the baggage of silly logos and nasty table layouts. In fact, my only quibble with RSS for news sites is that there's no predictable way to obtain a time and date for an article mentioned in an RSS feed, thus making it hard to unify several sources into one ordered list.
LiveJournal now supports RSS in the sense that peoples' journals can be syndicated in the same way. This is all well and good, and will be useful to some, but I have strong reservations for its usefulness as a long-term solution to syndicating journal entries between sites.
'Weblog' entries have a whole lot more metadata associated with them than your average news article, and that metadata is generally of more significance than it would be when attached to the current news headlines. Most importantly, we need to know where an entry came from, when it was generated (in a predictable format) and some other things which vary depending on what is being used to produce and manage the weblog.
For this reason, I think we should define an alternative XML application for syndication of weblog entries which is general enough that it could be used for any journal site, and push it as a standard which other sites can implement, but which mainly we can use for syndication between sites based on LiveJournal's code.
We could, if desired, extend RSS (which is, I believe, itself an application of RDF), but I see no reason in particular why that would be beneficial since there's no reason why we can't continue to support RSS too. In fact, RSS syndication onto friends pages will become very useful for the addition of news headlines onto friends views, as well as being able to push them into portal boxes. (which will, incidentally, be able to be incorporated into journal styles when S2 arrives)
A general XML format for weblog entries needs to separate several types of data to a greater extent than is possible with RSS. In particular, fields will be needed for a 'weblog name' (for example, "Mart's LiveJournal"), the URL to the weblog's 'recent entries', an optional subject or title line, the entry itself, a suggested small image to be displayed on sites which do such a thing, arbitrary metadata (which for LiveJournal would be the 'current' information, but specified in a general enough way that it could be used for anything that would be displayed in a similar way) and URLs for viewing the entry, adding comments and viewing comments.
I think that all of these items should be given on a per-entry basis to make it possible to make a friends view outputtable in this format, with entries from multiple sources. To save sending lots of redundant data when only viewing entries from one journal, similar values for 'source name' and 'source url' could be associated with the entire output and used as a 'default' when the 'weblog name' and 'weblog url' are not provided for the entry.
I'm not going to attempt to write an example now, because that would be better done after some discussion. The issue of getting other sites to implement it does of course remain, but at least all LiveJournal-based sites would support it, and there would still be our RSS support for the sites which are foolish enough to not implement our obviously superior solution! ;)