Behind the scenes of the Open Platform’s evolution

When I came to the Guardian two years ago, I brought with me some crazy California talk about open strategies and APIs and platforms. Little did I know the Guardian already understood openness. It’s part of its DNA. It just needed new ways of working in an open world.

Last week, The Guardian’s approach to openness and mutualisation took a giant step forward when we brought the Open Platform out of Beta.

It’s a whole new business model with a new technology infrastructure that is already accelerating our ambitions.

I’ll explain how we got to this point, but let me clarify what we just announced:

  • We’ve implemented a tiered access model that I think is a first in this space. We have a simple way to work with anyone who wants to work with us, from hobbyist to large-scale service provider and everything in between.
  • We’ve created a new type of ad network with 24/7 Real Media’s Open AdStream, one where the ads travel with the content that we make available for partners to reuse.
  • That ad network is going to benefit from another first which is Omniture analytics code that travels with the content, as well.
  • License terms that encourage people to add value are rare. Using many of the open license principles we developed T&Cs that will fuel new business, not stop it.
  • Hosted in the cloud on Amazon EC2 the service scales massively. There are no limits to the number of customers we can serve.
  • The API uses the open source search platform Solr which makes it incredibly fast, robust, and easy for us to iterate quickly.
  • We introduced a new service for building apps on our network called MicroApps. Partners can create pages and fully functional applications on

We’re using all the tools in the Open Platform for many of our own products, including the Guardian iPad app, several digital products and more and more news apps that require quick turn-around times and high performance levels.

Open Platform: Build applications with the GuardianThere’s lots of documentation on the Open Platform web site explaining all this and more, but I figured I could use this space to give a picture of what’s been happening behind the scenes to get to this point.

It’s worth noting that this is far from the full picture of all the amazing stuff that has been happening at the Guardian the past 12 months. These are the things that I’ve had the pleasure of being close to.

Beginning with Beta

First, we launched in Beta last year. We wanted to build some excitement around it via the people who would use it first. So, we unveiled it at a launch event in our building to some of the smartest and most influential London developers and tech press.

We were resolute in our strategy, but when you release something with unknown outcomes and a clear path to chaos people get uneasy. So, we created just large enough hurdles to keep it from exploding but a wide enough berth for those who used it to take it to its extreme and to demonstrate its value.

It worked. Developers got it right away and praised us for it. They immediately started building things using it (see the app gallery). All good signs.

Socializing the message

We ran a Guardian Hack Day and started hosting and sponsoring developer events, including BarCamp, Rewired State, FOWA, dConstruct, djugl, Music Hack Day, ScaleCamp, etc.

Next, we knew the message had to reach their bosses soon, and their bosses’ bosses. So, we aimed right for the top.

Industry events can be useful ways to build relationships, but Internet events have been really lacking meaning. People who care about how the Internet is changing the world and who are also actively making that change happen were the types of people we needed to build a long term dialog with.

So, we came up with a new kind of event: Activate Summit.

The quality of the speakers and attendees at Activate was incredible. Because of those people the event has now turned into something much more amazing than what we initially conceived.

Nick Bostrom’s darkly humorous analysis of the likelihood of human extinction as a result of technology haunts me frequently still, but the event also celebrates some brilliant ways technology is making life better. I think we successfully tapped into some kind of shared consciousness about why people invest their careers into the Internet movement…it’s about making a difference.

Developers, developers, developers!

Gordon Brown was wise in his decision to put Tim Berners-Lee and Nigel Shadbolt on the task of opening government data. But they knew enough to know that they didn’t know how to engage developers. Where did they turn for help? The Guardian!

We couldn’t have been more excited to help them get out the door successfully. It was core to what we’re about. As Free Our Data champion Charles Arthur joked on the way to the launch presentation, “nice of them to throw a party for me.”

We gave them a platform to launch in the form of developer outreach, advice, support, event logistics, a nice building, etc., but, again, the people involved made the whole effort much more impressive than any contribution we made to it.

Tom Taylor’s Postcode Paper, for example, was just brilliant on so many levels. The message for why open government data could not have been clearer.

Election data

Then when the UK election started to pick up some momentum, we opened up the Guardian’s deep politics database and gave it a free-to-use API. We knew we couldn’t possibly cover every angle of the election and hoped that others could use the Guardian’s resources to engage voters. We couldn’t have asked for a better example of that then Voter Power.

A range of revenue models

All along there were some interesting things happening more behind the scenes, too.

The commercial team was experimenting with some new types of deals. Our ad network business grew substantially, and we added a Food Ad Network and a Diversity Network to our already successful Green Ad network.

It was clear that there was also room for a new type of ad network, a broader content-targeted ad network. And better yet, if we could learn about what happens with content out across the web then we might have the beginnings of a very intelligent targeting engine, too.

24/7 Real Media’s Open Ad Stream and Omniture were ready to help us make this happen. So, we embedded ads and analytics code with article content in the Content API. We’ve launched with some house ads to test it out, but we’re very excited by the possibilities when the network grows.

The Guardian’s commercial teams, including Matt Gilbert, Steve Wing, Dan Hedley and Torsten de Reise, also worked out a range of different partnerships with several Beta customers including syndication, rev share on paid apps, and rev share on advertising. We’re scaling those models and working out some new ones, as well.

It became obvious to everyone that we were on to something with a ton of potential.

Rewriting the API for scale

Similarly, the technology team was busily rebuilding the Content API the moment we realized how big it needed to be.

In addition to supporting commercial partners, we wanted to use it for our own development. The new API had to scale massively, it had to be fast, it had to be reliable, it had to be easy to use, and it had to be cheap. We used the open source search platform Solr hosted on Amazon’s EC2. API service management was handled by Mashery.

The project has hit the desk of nearly every member of the development team at one point or another. Here are some of the key contributions. Mat Wall architected it. Graham Tackley made Mat’s ideas actually work. Graham and Stephen Wells led the development, while Francis Rhys-Jones and Daithi O’Crualaoich wrote most of the functions and features for it. Martyn Inglis and Grant Klopper handled the ad integration. The wonderful API Explorer was written by Francis, Thibault Sacreste and Ken Lim. Matthew O’Brien wrote the Politics API. The MicroApps framework included all these people plus basically the entire team.

Stephen Dunn and Graham Tackley provided more detail in a presentation to the open source community in Prague at Lucid Imagination’s Solr/Lucene EuroCon event.

The application platform we call MicroApps

Perhaps even more groundbreaking than all this is the MicroApp framework. A newspaper web site that can run 3rd party apps? Yes!

MicroApps makes the relationship between the Guardian and the Internet feel like a two-way, read-write, permeable membrane rather than a broadcast tower. It’s a very tangible technology answer to the openness vision.

You can learn more by reading 2 excellent blog posts about MicroApps. Dan Catt explains how he used MicroApps for Zeitgeist. Since most of the MicroApps that exist today are hosted on Google AppEngine, the Google Code team published Chris Thorpe’s insights about what we’re doing with MicroApps on their blog.

The MicroApps idea was born out of a requirement to release smaller chunks of more independent functionality without affecting the core platform….hence the name “MicroApps”. Like many technology breakthroughs, the thing it was intended to do becomes only a small part of the new world it opens up.

Bringing it all together

At the same time our lead software architect Mat Wall was formulating the MicroApp framework, the strategy for openness was forming our positioning and our approach to platforms:

…to weave the Guardian into the fabric of the Internet; to become ‘of‘ the Web, not just ‘on‘ the Web

The Content API is a great way to Open Out and make the Guardian meaningful in multiple environments. But we also knew that we had to find a way to Open In, or to allow relevant and interesting things going on across the Internet to be integrated sensibly within

Similarly, the commercial team was looking to experiment with several media partners who are all thinking about engagement in new ways. What better way to engage 36M users than to offer fully functional apps directly on our domain?

The strategy, technology and business joined up perfectly. A tiered business model was born.

The model

Simon Willison was championing a lightweight keyless access level from the day we launched the Beta API. We tested keyless access with the Politics API, and we liked it a lot. So, that became the first access tier: Keyless.

We offered full content with embedded ads and analytics code in the next access level. We knew getting API keys was a pain. So, we approved keys automatically on signup. That defined the second tier: Approved.

Lastly, we combined unfettered access to all the content in our platform with the MicroApp framework for building apps on the Guardian network. We made this deep integration level available exclusively for people who will find ways to make money with us. That’s the 3rd tier: Bespoke. It’s essentially the same as working in the building with our dev team.

We weren’t precisely clear on how we’d join these things up when we conceived the model. Not surprisingly, as we’ve seen over and over with this whole effort, our partners are the ones who are turning the ideas into reality. Mashery was already working on API access levels, and suddenly the last of our problems went away.

The tiers gave some tangible structure to our partner strategy. The model felt like it just started to create itself.

Now we have lots of big triangle diagrams (see below) and grids and magic quadrants and things that we can put into presentation slides that help us understand and communicate how the ecosystem works.

Officially opening for business

Given the important commercial positioning now, we decided that the launch event had to focus first and foremost on our media partners. We invited media agencies and clients into our offices. Adam Freeman and Mike Bracken opened the presentation. Matt Gilbert then delivered the announcement and gave David Fisher a chance to walk through a deep dive case study on the Enjoy England campaign.

There was one very interesting twist on the usual launch event idea which was a ‘Developer Challenge’. Several members of the development team spent the next 24 hours answering briefs given to us by the media partners at the event. It was run very much like a typical hack day, but the hacks were inspired by the ideas our partners are thinking about. Developer advocate Michael Brunton-Spall wrote up the results if you want to see what people built.

Here is the presentation we gave at the launch event:

(Had we chosen a day to launch other than the same day that Google threw a press release party I think you’d already know all this.)

Do the right thing

Of all the things that make this initiative as successful as it is, the thing that strikes me most is how engaged and supportive the executive team is. Alan Rusbridger, Carolyn McCall, Tim Brooks, Derek Gannon, Emily Bell, Mike and Adam, to name a few, are enthusiastic sponsors because this is the right thing to do.

They created a healthy environment for this project to exist and let everyone work out what it meant and how to do it together.

Alan articulated what we’re trying do to in the Cudlipp lecture earlier this year. Among other things, Alan’s framework is an understanding that our abilities as a major media brand and those of the people formerly known as the audience are stronger when unified than they are when applied separately.

Most importantly, we can afford to venture into open models like this one because we are owned by the Scott Trust, not an individual or shareholders. The organization wants us to support journalism and a free press.

“The Trust was created in 1936 to safeguard the journalistic freedom and liberal values of the Guardian. Its core purpose is to preserve the financial and editorial independence of the Guardian in perpetuity, while its subsidiary aims are to champion its principles and to promote freedom of the press in the UK and abroad.”

The Open Platform launch was a big day for me and my colleagues. It was a big day for the future of the Guardian. I hope people also see that it was a major milestone toward a brighter future for journalism itself.

One thought on “Behind the scenes of the Open Platform’s evolution”

Comments are closed.