Home > Archive >  2009 >  September >  25

Where we're at with rssCloud

Friday, September 25, 2009 by Dave Winer.

A brief report on where we're at with rssCloudLand as September winds down.  

It's been a great summer. In July, we got busy rebooting the RSS cloud, based on the experience in 2001 and 2002, with lessons learned, and a lot more success than we had seven years ago. 

Based on the walkthrough document, the community has shipped multiple implementations of all three sides of the interface. 

1. There have been many implementations of the cloud software, including my own rssCloud.root and wordpress.com and the plug-in that turns every standalong WordPress installation into an rssCloud server. Total number of installations: millions.  

2. Lots of feeds and authoring systems numbering in the millions. Example real-time podcast feeds and photo feeds. 

3. Not so many aggregators. So far only two -- first my own River2 and LazyFeed. There many more on the way. I hope one or two Twitter clients will ship rssCloud support in the month of October.  

What next?  

I have four immediate priorities: 

1. Features that will make it possible for Google Reader and other large scale aggregators to wire into rssCloud. These are technical features, to most users they are details, but nonetheless are important. It must be possible for a server to register a notification request using a domain name, on behalf of another server. And along with this feature, most developers believe a stronger verification process is needed. I am one of them. Yet it must remain simple to implement, so the barrier to entry is as low as possible. There have been proposals in both these areas.  

2. Long-polling proxies for clients running on systems behind NATs or firewalls. This technology is well-understood, and a number of developers are working on solutions. I hope when code emerges, there will be an effort to make their interfaces compatible, so maximum interop can be achieved. 

3. Conventions for naming, so that URLs can be mapped to names that are managed by DNS, the naming system of the Internet itself. I've done a simple experiment in this area to get myself educated on the issues and learn how users and developers see this.  

4. A high-level user-oriented vision statement for rssCloud with a roadmap for developers and marketers. My goal is to create an open loosely-coupled message network that achieves more or less what Twitter does, but in a decentralized manner. It must use the architecture of the Internet to achieve what Twitter achieves in a centralized fashion. No doubt there will be tradeoffs, some things the open network will do better, and other things that the centralized one will. This is similar to previous layers of the Internet, where systems like AOL and CompuServe provided more sophisticated features, where the World Wide Web was more low-tech, more basic. Both ways of doing things have advantages and survive to this day. 

Comments, questions and suggestions are welcome.