Part 3 of the The RSS Diaries: Interviews about Syndication in the Enterprise.
July 13, 2005.

inkBlots: As one of the co-founders of the open-source ROME project, you have considerable amounts of time and energy vested in handling syndication formats. What drove you to go from a simple user of RSS and Atom feeds to start a project dedicated to supporting them?

Patrick Chanezon: My first experience with RSS was not as a user, but supporting infrastructure to deal with it on the server side. RSS was invented by Dan Libby at Netscape in 1999 in order to let users syndicate their content on the My Netscape portal. I joined the NetCenter divlet soon after that, and after a few months (after Dan left), became the engineering manager for the My Netscape team. The RSS fetcher infrastructure was part of my team’s responsibilities.

Later on I switched from AOL to Sun and dealt with other types of applications. In 2002 I started using Radio UserLand and discovered weblogs and client side RSS aggregators. I found blogging a very interesting activity, but had a hunch that syndication, not blogging, was the killer app. Using the rich client RSS reader NetNewsWire was an aha moment for me: it allowed me to follow 300 feeds instead of the 10 tabs I opened daily in Mozilla at that time. Using this new application made me feel as enthusiastic as when I first used a mail client and later a browser. This further convinced me that syndication formats had a potential disruptive effect.

In 2003 I started evangelizing blogs, wikis and RSS at Sun, in order to create products in that space. In 2004 Jonathan Schwartz announced in a eWeek article that Sun was adopting RSS, just before he was named president of Sun.

During that period, Alejandro Abdelnur, Elaine Chien and myself started this internal project for a common syndication parsing and generation library: we thought that after Jonathan’s public stance many groups at Sun would start creating syndication apps and it made sense to all use the same library.

It’s at that time that we thought there was no competitive advantage in this library, and that by making it open source we could leverage the efforts and knowledge of a much wider developer community. We created the ROME open source project on java.net in a day.

Interest grew quickly around ROME, so that today we have more than 12 developers and many projects using it. See the Powered By ROME Wiki page.

iB: Who are the typical users of ROME and what problems are they trying to solve?

PC: In general our users are java developers who create applications dealing with feeds. We haven’t done an extensive study about who uses ROME for what, so all we have as an indicator is people who added their project to the Powered By ROME Wiki page, and questions asked in the users and developers mailing lists. I categorize our users along 2 axis: open source vs closed source developers, and type of applications.

There are a few open source projects using ROME, from Wikis (xWiki, SnipSnap) to Dave Johnson’s new Roller Planet aggregator. ScheduleWorld, a Calendaring application, FeedPod, which uses a speech to text library to add sound enclosures to your textual feed.

And there are commercial offerings, such as Public Interactive, which provides podcasting capabilities to Radios in the US, and Reger.com, where Joe Reger uses ROME to explore new grounds in what he calls Data Blogging.

iB: How is Sun Microsystems using syndicated feeds within its own enterprise to improve business?

PC: The first useful application related to feeds that was used at Sun was Planet Sun: Dave Edmondson created this aggregator of all Sun blog external feeds (those from Roller and those from Sun people blogging on their own system) soon after we launched blogs.sun.com. Before that Simon Phipps, Sun’s chief evangelist, had a page where he listed the few Sun external bloggers (himself had been blogging for years). When the number of Sun bloggers exploded, Dave Edmondson setup the Planet software (a Python based aggregator) and started aggregating them all. He also setup an internal Planet for internal blogs: these are really useful to track what Sun people are saying.

Later on, Dave started creating topical aggregators such as PlanetSolaris for Sun and non-Sun people talking about Solaris, and Planet WallStreet, which is a hack I suggested to Hal Jespersen: aggregate Roller posts based on category name. Thus in order to create a team blogs about a specific topic, you just have to agree on using the same category name.

Dave Johnson, from Roller fame, who’s now working at Sun, recently started developing the same functionalities inside Roller, based on ROME and ROME Fetcher: it’s called Roller Planet. ROME was started to make creating such an online aggregator easy.

Eduardo Pelegri-Llopart also started some interesting use of syndication on the java.net JWSDP community, of which ROME is part.

It took a year for Sun employees to become familiar with blogging, and now [they are] a leader in corporate blogging. Armed with this blogging experience, Sun employees start to understand the power of syndication and define some requirements for how to use it to improve business. Look forward to more innovation from Sun in this area in the next year.

iB: Based on your experience at Sun Microsystems, do you see any obstacles that slow the adoption of syndicated feeds within corporate infrastructure? How do you think these obstacles can be overcome?

PC: The successful uses of syndicated feeds to improve business that I listed above were all ad hoc innovations started by individuals on their own servers, then adopted by everybody. I think this is the way innovation happens in companies today. The adoption of blogs was different: we had a very strong will from the top (Jonathan Schwartz, Tim Bray) to get started, and a very strong Will (Snow, the guy in charge of sun.com) to make it happen in a week.

As an example of what happens when internal IT is involved, and no strong orders come from the top, 9 months ago I created a better RSS channel based on ROME for Sun’s internal Portal. Today it is not deployed yet. I don’t blame them, they have security, scalability and performance requirements for official internal infrastructure that more ad hoc projects don’t have (the internal blog provisioning application that I had created for blogs.sun.com was a quick hack full of bugs, which I mainly developed live on the production server, but it was OK because speed of deployment was the main design criteria).

“The biggest obstacle to adoption of a new technology is the fact that IT departments are always slower than end users to deploy them.”The biggest obstacle to adoption of a new technology is the fact that IT departments are always slower than end users to deploy them. Jonathan blogged about that in “Who picked your search engine?.” I won’t debate in this post about how to make IT departments more agile, since this is not the topic at hand, and I would have too much to say about it. It is just a fact of current corporate life that you have to take into account.

In the case of syndication application, there are plenty of free desktop aggregators on all platforms, so end users will adopt them quickly. But for more innovative and sophisticated server side applications like the ones you find on the public Internet (Bloglines, rojo, feedburner, delicious), adopting them internally is more challenging.

iB: Do you see any trends in the use of RSS and Atom? What changes do you see in the next year or two for syndicated feeds in the marketplace?

PC: I’ve spoken a bit about that in my blog entry about the typology of syndication applications, and in my recent XTech 2005 ROME presentation. I’ll talk about it in our common session at JavaOne June 30th Beyond Blogging: Feed Syndication and Publishing With Java™ Technology

To summarize, what I see coming is:

• Desktop syndication suites, integrating a blog poster and a syndication reader, like Radio UserLand in the olden days, and NetNewsWire/Marsedit today on the Mac.
• More integration of syndication functionality in the browser and mail clients: this has happened already for FireFox and the latest Safari, as well as Thunderbird, and OutLook with the NewsGator plugin. I bet that it won’t be long before Microsoft adds these functionalities in IE and OutLook.
• Server syndication engines: current online syndication servers are just a sink for feeds, that you access in a classic web application view. I think this is not where the best potential lies for syndication applications. A new class of server side applications will appear, that I call server syndication engines, which are both sinks and sources of syndication data. These will provide functionality to let you filter, augment, and remix your feeds, or send them in business process engines. Early implementations of these concepts can be found in Merkaat, Feedster, Feedburner, PubSub and AllConsuming.
• Syndication formats as an alternative envelope format (alternative to SOAP) for time based results of REST style web services calls. RSS started as a content aggregation technology but will evolve in an application integration envelope format, with people embedding namespaced application specific meta data in an RSS or Atom envelope. Amazon a9 opensearch extension to RSS 2.0 is a good example of that. But I think things like UBL may end up being embedded in a syndication envelope for B2B applications data interchange. Joe Reger with his Structured Blogging experiments (last year he started adding Marathon related meta data in his feed), and Bob Wyman’s Data Blogging proposals are hints that this will happen sooner than later. ROME modules were created with this type of use case in mind.
• Desktop aggregator plugin architecture: some desktop aggregators, whether in browser or not, will develop some kind of plugin architecture in order to display and let the user interact with these data items embedded in feeds.
• Tagging: feed management and tagging are a marriage made in heaven. Rojo are the first ones to have spotted that trend. I expect this to become a big trend.
• Atom and the Atom publishing protocol will gain adoption in many content related servers: content management systems, Wikis (xWiki is working on it)

iB: What advice do you have for the small business or department head who is looking to incorporate syndicated feeds into their business processes?

PC: Start simple, work iteratively, XP style. Stay away from discussing this with your IT department, or making big infrastructure decisions.

Start with a quick prototype, using either ROME if it’s in java, or UFP if it’s in Python. Buy Dave Johnson’s Syndication in Action book when it comes out: it covers everything, including the Atom Pub protocol. You could also start with Roller Planet or Planet, and tailor them to your needs. See if it gains adoption amongst users.

If so, and you have more ambitious applications in mind, start looking at what startups propose in terms of syndication servers.

Today I’m only aware of 2 companies that have an offering in this space: NewsGator Enterprise, and Hit Syndicaat RSS Enterprise Syndication Management System.

But I’m sure there will be many more in the next year.

iB: Will the world of syndicated feeds play a part in your new job at Google?

PC: I would say 80/20.  

80% No. My role at Google is to be in the Adwords API developer relations. Make sure developers using the API are happy, try to grow a community around the API.

20% yes. At Google we have this notion of 20% project, where you can spend 20% of your time on a project of your choosing (it can be a Google project or an open source project). For details [of this notion] see Joe Beda’s post. Considering my passion for syndication feeds, you can bet I’ll spend my 20% project on something related.

Patrick would also like to note that the opinions he expresses are his own and not necessarily that of his employers, past or present.

Return to The RSS Diaries: Interviews about Syndication in the Enterprise.