aaron crews online

Twitter and the cloud

A few months ago Gary Mintchell pointed me to this post from Dave Winer, where he proposes some changes to twitter.  Essentially, he says there's value in twitter as a protocol, but that it should be distributed and made available for people to use on their own individual websites.  Twitter conversations could spring up on individual sites around certain topics or posts.  That allows the individual sites to harbor community and to hold the attention of their readers.

I've seen a lot of new technology lately that's also designed to do just that - to bring the conversation around the web back to the originating sites.   Friendfeed provides rooms and widgets to bring it back somewhat, Disqus has a "Reactions" section where the web is scanned for conversation surrounding the post, and most recently JS-Kit introduced Echo, which is designed to aggregate discussion of content in real time on the originating site.  Of course this all is pretty slow or difficult to set up - by its nature you have to use search and/or have to poll a variety of sites/feeds.  Combining group activity on twitter has been limited to the use of hashtags, which take up characters, require a consensus, and take effort/memory to use.

Dave later highlighted Automattic's P2 as a start on a distributed "twitter" publishing tool.  As a matter of fact, after seeing his tweet on the subject I was inspired to create a local P2 instance for our company to use inside our firewall.  There are some architectural issues with that, though.  How do you group related posts for small teams?  If you create seperate P2 instances for each team, how do you share knowledge between them?  So far the only solution I could see was some kind of inside-the-firewall rss aggregator.  As a test solution, I put in an instance of tt-rss, which was ok, but not really multiuser.  Plus, you have other issues like multiple logins, maintenance headaches, different interfaces, etc.

The issue I was seeing with twitter as a protocol is that it was essentially just a handicapped (140 character) version of multiuser blogging or group chat (or commenting even).  What about it would make it "twitter?"  I suggested that all these distributed instances would push content to a twitter cloud, which could be used for trends/aggregation/live search.

Little did I know, Mr. Winer was baiting me and had something up his sleeve.  Today I saw him link to a new project, http://rsscloud.org, which essentially combines these three ideas: (1) a twitter (140-character format) publishing tool which pushes content to (2) a cloud which can be polled by (3) aggregators.  A source tag ties everything back to its original content, so aggregation is easy.

It's something that apparently existed all the way back in 2001, but only amongst Radio UserLand.  If twitter catches on as a protocol, I think this is a truly powerful concept.  I wonder how it all fits in with the activity stream standard and FriendFeed's SUP proposal and all the other needs out there.