Part 2 of the The RSS Diaries: Interviews about Syndication in the Enterprise.
April 17, 2005.

inkBlots: Many people know you from your contributions on XML-DEV and your recent book on RSS and Atom. How did syndicated feeds become such a deep interest for you?

Danny Ayers: Mostly selfish reasons. Although I was aware of RSS from early on, I didn’t really see the potential until blogging took off and I wanted to check more than a dozen sites on a daily basis. Then there was the fact that this material is machine-processable data, the kind of thing needed to make the Web more useful. I like it when the computers do more of the work.

iB: The back cover of your book seems primarily aimed at bloggers. Was the idea of syndication in the enterprise too premature at the time you went to press, or were there other factors involved?

DA: Heh, I was happy to leave the back of the book to the publisher. The emphasis in the book is very much on the public Web environment, and right now bloggers are a key part of that. But you’re right though, syndication in the enterprise is a very significant aspect of these technologies. But I’m not sure whether more emphasis on that would make much difference, historically enterprise people have picked up on what’s happening on the Web at large sooner or later. The book’s target audience is really just programmers, the techniques being generally the same either side of the company firewall.

iB: Many software and consulting companies like NewsGator and MyST
Technology Partners are actively targeting customers for enterprise-wide publishing and syndication frameworks. What do you see as the driving forces behind this new-found adoption of RSS and Atom in the business world?

DA: That would be the enterprise people picking up on what’s happening on the Web at large . Primarily it provides another interface for producing, distributing and consuming data, better suited to a lot of tasks than traditional document/content management or email-style communications. The time axis is very important in the enterprise, and the reverse-chronological delivery of feeds fits this nicely. But that doesn’t really answer the question. The biggest driving force is probably the growth of blogging, in particular those high-profile bloggers respected in the IT industry.

iB: As an evangelist of the Semantic Web, you have also become a big proponent of the Atom specification. What will Atom do for the Semantic Web that RSS 1.0 has failed to do?

DA: Evangelist? “Tubthumper” might be closer to the mark… Anyhow, the biggest failings of RSS 1.0 are those of public relations and marketing. The net result is that feeds in general are far poorer quality than they need have been. Atom offers the potential for things to get cleaned up, and for a very high proportion of syndicated data to be good quality. This is good for the Web, and anything that is good for the Web is good for the Semantic Web. Don’t forget the base layers of the Semantic Web stack are HTTP, URIs and XML - the building blocks of Atom. That’s the short answer.

“I suspect RSS 1.0 was the right idea at the wrong time.”Long answer. I suspect RSS 1.0 was the right idea at the wrong time. On the one hand you had the unproven benefits of the syndication approach to Web publishing, on the other the unproven benefits of Semantic Web technologies. This was at a time when most people were only just beginning to get to grips with XML. Namespaces? Horror. Moving on from there, most syndication toolbuilders have worked directly against the single-purpose feed/content/metadata kind of model, and from that viewpoint RSS 1.0 and RSS 2.0 are pretty indistiguishable. Right now the systems that exploit the RDF side of RSS 1.0 are in a minority. Oh, and “2.0″ sounds shinier than “1.0″. However, Semantic Web technologies are very good for information integration (by design), and the RDF approach is still workable even if the original data is found in domain-specific XML (or for that matter relational databases). If you can map between the existing model and the RDF model, you can integrate (XSLT is one route). In principle at least RSS 2.0 is mappable. But being able to translate coherently means the source data must be coherent. The ambiguity of content escaping is an annoying flaw frozen into RSS 2.0, but a wider point is that the promotion of RSS as “simple” has led to huge variability of data in the wild. Simple is usually good, but ignoring good practice leads to problems down the line. Atom is designed to be simple, but also to do things *properly*. This makes life for the developer much easier as the data’s more predictable. From the Semantic Web point of view, even though Atom isn’t RDF-based it will be possible to do reliable conversions, because there’s a solid spec to work from.

Even if you discount all these problems as details, the fact is both RSS 1.0 and RSS 2.0 (as a direct descendent of 0.91) were designed years ago, before the explosion in blogging and other significant changes in the Web landscape. It’s a different environment, with different demands. Atom is designed for those demands. The Atom Protocol will hopefully fill a gap in content publishing which is currently covered by less Web-native approaches like ftp and XML-RPC. Again this is good for the Web, hence good for the Semantic Web.

Incidentally, it’s worth noting that RSS 1.0 is far from dead. Most producers and consumers support both RSS 1.0 and 2.0. One benefit the RSS/RDF approach does offer is a controlled approach to extensions that lie outside of the core RSS vocabulary. For example, people like Nature Publishing add extended bibliographical information to their feeds. In a straight XML format this would mean throwing in arbitrary namespaced elements into their feed and hoping everyone that consumed that data would build understanding of the relationships between the extra material and the core data. It’s not only hard work but the outcome is unpredictable. In RSS 1.0 it’s not left to chance, because those relationships are expressed in the RDF model. Last thing I heard Nature were reluctant to move to Atom because it didn’t have as manageable an extension model as RSS 1.0.

iB: RSS 1.0 is based on the RDF specification, yet RDF has a reputation for complexity. How do you answer the frequent “RDF is too hard to manage” cry from developers?

DA: Try it. People familiar with XML tend to focus on RDF/XML, the format of RSS 1.0, which looks complex compared to some many XML formats. But that’s missing the point. The model is the important part, and that’s essentially very simple (as is the Turtle syntax, incidentally). Sure, in one-off, domain-specific applications it can be a lot easier to use just XML or a relational database with custom schemas, custom models. But those lead to very inflexible solutions. If you need to change the data structure or integrate with other systems, having a simple common model can make the system considerably easier to manage.

iB: What are the short-term and long-term benefits a business can realize by using Semantic Web-friendly syndication within their corporate infrastructure?

DA: It’s mostly about maximising the utility of the data you’ve got. If enterprise-wide data is integrated then it can give you better business knowledge. Syndication can offer a convenient means of keeping information current, keeping people up-to-date, across an enterprise. Regarding the Semantic Web angle, short-term you get the immediate benefit of being able to integrate information in a fashion that more naturally follows the structure of the information. Basically you don’t have to cram everything into tables or trees. Mid-term you have a system that’s inherently better at supporting change, including extension into areas that weren’t originally planned. Hopefully long-term you’ll have the enormous network-effect benefits of being part of a Semantic Web.

iB: What is your advice to the small business who wants to try using RSS or Atom to improve their business processes? How should they get started?

DA: In a word: Blog! Start blogging and using an aggregator yourself, and encourage your staff to do likewise. any blogging tools and aggregators are freely downloadable, open source and easy to set up (e.g. WordPress, RSS Bandit). Even if your own blog is only on the Intranet, reporting deliveries of office stationery, it’ll give you an idea of what syndication’s good for. I suspect every different company will have different ways in which they can improve their business processes with these technologies, trying out the typical blog/syndicate setup should help you identify the best leverage points. A travel agency might announce last-minute offers direct to the public, a double-glazing company might deliver positive thinking podcasts to its sales staff in the field. Don’t underestimate the potential of blogging publicly and aggregating from the Web. A blog can do huge amounts for raising awareness of your business, and provide a human face that people can connect with. An aggregator can kept you current with events in your industry. There is also huge potential for syndicated data distribution where the immediate data creator and/or consumer aren’t human. Pulling something from the sky - how about data syndicated from weather stations being used to turn off crop sprinklers when rain was on its way. Oh yes, and I would of course also suggest you read up on Semantic Web technologies, in these early days there’s the potential of considerable commercial advantage for early adopters that play their cards right.

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