A Tale of Agile Development… of a Standard

By   /  December 14, 2012  /  No Comments

Some in the Semantic Technology community have pointed out that from a development perspective, Semantic Technologies are well suited for an agile approach to programming, and we will be discussing that idea more in future here at SemanticWeb.com. Today, however, we’re taking a look at some novel thoughts on agile development of a standard, thanks to guest contributor, Andreas Gebhard. He is Director, Editorial at Getty Images, and Board member of the IPTC.

We caught up with Gebhard at the recent Semantic Technology & Business Conference in New York, where he initially shared this idea with us.

He has expanded on these ideas in a post on the Getty Images blog. As Gebhard says, “I want to tell you the story of how we got there in just about a year — tremendously fast, in the world of standards.”

We re-print the post in its entirety below with thanks to the author and Getty Images.

IPTC rNews — An example of agile standard development

by Andreas Gebhard on Dec 13, 2012

rNews is a new standard for embedding semantic, machine-readable metadata in web documents [1], but the standard itself is not what I want to talk about. Instead, I want to tell you the story of how we got there in just about a year — tremendously fast, in the world of standards.

The term “agile,” referring to software development, has become a buzzword in the technology community since its introduction in 2001, but I have never before seen it applied to standards development. According to Wikipedia,

Agile software development is […] based on iterative and incremental development, where requirements and solutions evolve […]. It promotes adaptive planning, evolutionary development and delivery […] and encourages rapid and flexible response to change.

Traditionally, work in standards bodies tends to follow the rules of the so-called waterfall model instead. Waterfall methodology promotes thorough up-front research, thorough discussion and feature definition and finally, the design and construction process.

Suffice to say, the two models are not compatible with one another: Any sizable project takes years to materialize under the waterfall model. When the IPTC [2]formed a working group to develop rNews in 2010, we agreed on three guiding principles:

  1. rNews had to be simple to implement
  2. the rNews schema would focus only on publishable metadata (i.e., not contain behind-the-scenes and workflow-related properties)
  3. rNews had to be ready soon

Simplicity. Focus. Speed.

The third requirement was somewhat unusual, compared with much of the previous work the IPTC (or most other standards bodies, for that matter) had done: In standards development, there are rarely deadlines. A standard is ready when it’s ready.

Obviously, people try to finish their work as soon as they can but if a detail requires further discussion, the necessary time is usually taken, because a standard, once approved and published for the world to use, must be reliable. It must allow people and organizations to base their products, processes and projects on without the fear of having to change everything a few months down the road: It’s simply an issue of trust.

Simplicity. Focus. Speed. Trust.

It’s only apparent in hindsight that in order to achieve all of that, we had to employ many of the methodologies that have come to be known as “agile” — but we did so organically, without an upfront decision or discussion.

To keep it simple, we started with our experience in the industry. We made assumptions. We discussed them, agreed and wrote drafts. We actively sought feedback. We incorporated new ideas, threw old ones overboard, disagreed, discussed some more and made compromises.

We wrote more drafts. We tried implementing our own ideas, ran into walls or little stumbling blocks. We asked for advice, received valuable help[3] and went back to the drawing board again.

Our four guiding principles focused us: we could test any new idea or assumption against them. If a given idea did not further one of these principles or, worse, contradicted it, we killed the idea.

Our speed is also owed to a bit of a coincidence: the core authors behind rNews are only three people[4] and we are all based in the same area. One of the drawbacks of that was that it turned out to be a lot of work for each one of us (in addition to our respective day jobs). But the huge benefit was that we could all quickly join calls, meet in person and break ties when we disagreed.

We gained the necessary trust by doing all this very publicly. We talked to people, spoke at meet-ups and conferences, blogged and tweeted about it. We were (and continue to be) approachable and we stand behind the work we’ve delivered.

Shortly before our release date we were approached by Google, Bing and Yahoo, the sponsors behind schema.org (later to be joined by Yandex). We wanted the search engines to endorse rNews. They wanted us to collaborate on schema.org. Both sides understood that a competition between two standards with very similar goals would be a waste of everyone’s energy.

The result were some pragmatic decisions and excellent collaboration on both sides. The IPTC’s Board of Directors even approved a change in license terms to accommodate the needs of schema.org, a fairly complex legal procedure for an established industry standards body. (Full Disclosure: I am a member of the aforementioned board.)

I think all involved, inside and outside the IPTC and its member organizations, can be rightly proud of how quickly rNews has come this far and how it has been made available to millions of web publishers, many of whom are implementing it via schema.org without any knowledge of the IPTC and its origins.

Two things I’d like to leave you with…

1. If you or your company builds tools and technology for the media or publishing industry or is itself a part of this industry, have a look at the IPTC. They don’t bite, they love to work with smart people and they get stuff done. If you have questions about joining, let me know and I’ll put you in touch.

2. The more important one: If you want to define a standard, know your audience and your goals. Define them so you can test against those definitions. Release “betas” and iterate. It doesn’t have to take years.

Questions? Comments? Leave a comment below or ask me on Twitter or app.net.


[1] It’s a mouthful, I know. It means that publishers who implement rNews can mark (in a machine-readable way, e.g. for search engines) things like the headline, authors’ and editors’ names, copyright and usage terms. Better yet, we added powerfully simple ways to mark up the people, organizations, locations and topics that the published text (or photo, video, audio) is about. For more about rNews itself, please check out our extensive documentation at http://rnews.org/

[2] The International Press Telecommunications Council is a global media industry standards body that was formed in 1965 and has since produced numerous standards that are used in the publishing industry and outside. The IPTC is probably most widely known for its Photo Metadata standards. More information is available at http://iptc.org or on Wikipedia (though slightly outdated).

[3] A few of the people who deserve recognition in this regard: Michael Steidl (Managing Director of the IPTC), Vincent Baby of Thomson Reuters, Jarred McGinnis of the Press Association (PA), Jean-Pierre Evain of the European Broadcasting Union (EBU), Jayson Lorenzen of Business Wire and also Raphaël Troncy of EURECOM.

[4] Stuart Myles of the Associated Press (AP), Evan Sandhaus of the New York Times and myself, representing Getty Images.


About the Author

Photo of Andreas GebhardAndreas Gebhard started 20 years ago as a photojournalist and worked at a number of publications and news agencies since. As Director, Editorial at Getty Images, he is responsible for operations of the New York news picture desk but focuses on data and technology projects. He represents Getty Images in the IPTC and most recently co-authored rNews, a new standard to embed publishing metadata in web pages in RDFa and Microdata. Gebhard is a native of Berlin, Germany, and lives in Brooklyn, NY.

You might also like...

Predictive Analytics Use Cases

Read More →