Deep down inside the Web

The World Wide Web Consortium (known universally as W3C) continues to be the driving force behind protocols and guidelines that ensure the long-term evolution and development of the Web. We took the opportunity of talking to one of the leaders behind that process, W3C Fellow Dave Raggett.

Dave Raggett has been deeply involved with the development of key Web standards since 1992. In 1994 he launched and chaired the IETF (Internet Engineering Task Force) HTTP working group and drove early standards work on HTML+, HTML 3.0, HTML tables and HTML forms. Since 1995, he has been a Fellow of the Boston-based World Wide Web Consortium (W3C), the body started, and still chaired by Sir Tim Berners-Lee, the inventor of the WWW.

We met up with Dave Raggett at home, grabbing a quiet moment in his dizzying schedule of international travel, to talk about his role in W3C. This includes work on standards for a new generation of authoring tools for distributed Web applications. They involve a wide diversity of devices such as desktop computers, office equipment, home media appliances, mobile devices, physical sensors and effectors (including RFID and bar-codes). It is self-evidently of interest to the air transport community.

SITA: What is driving this work?

DR: People are increasingly using the Web for applications, but html has become more complicated, which makes it harder to build applications. It is also increasingly expensive, because of variations across browsers - not just versions but also different vendors. So W3C is exploring the idea of using model-based approaches to support cheaper and easier authoring of applications.

Take the range of skills and people involved in producing a web application. Some are worried about graphic design, some about the business rules, others about the user experience. They must all be concerned with accessibility and privacy. A whole range of people are involved, each bringing different concerns and different perspectives. To be effective, they need to be able to work independently. But the current approaches of markup, scripting and so on don't really cut it, because it's all jumbled up together.

Model-based approaches allow for a separation of concerns via models that support the different roles and perspectives people bring to their work. The authoring tools look after the complexity involved in keeping the different levels of models in sync.

Another part of the Group's work is to help people adapt applications to a particular delivery channel, especially since now we have mobile devices and TVs that offer web-browsing technology.

SITA: The principle seems to be the same that has always underpinned the Web - keep it simple so that more and more people can do it.

DR: That's right. You can't really cope with all these challenges if you're buried in the code or the scripts, so you need to find a way of abstracting away from that. The W3C Ubiquitous Web Applications Working Group is focusing on context adaptation and personalization. It's about what model you have for the user preferences of the device capabilities and environment and then how can that be exposed to applications.

SITA: If we take mobiles, there's great variety in appliances that have greater or lesser levels of web access...

DR: Our Group did evolve originally from work being done on mobiles, but it's broadened out. There was a device independence working group but they were talking about the concept of device independent authoring, preceding the mobile web initiative. There are a number of other groups within W3C that focus on the mobile industry.

Some of the standards work is about taking material that's already been deployed, but where companies want an interoperable standard because customers are complaining... They reckon they'll grow the market if they build the standard. So standards groups can knock off the rough edges of an already de facto standard. Companies are willing to work together to do that.

Then there's work that is typically in the pre-competitive phase. We're trying to put mileposts down, as it were, around which research groups and industry can steer by. We're developing an ontology - essentially a model or vocabulary - that can be exposed into different interfaces. The idea is that, although there may be different interfaces or contexts, the work is all based on the same fundamental assumptions. An example would be the difference between portrait and landscape. What does that mean?

SITA: Your work is still based on those first days of Berners Lee trying to simplify things, making the Web accessible and available to the biggest number of people at the minimum cost?

DR: Yes. Essentially it's fulfilling the potential of the Web...

SITA: On the understanding that the Web is something ubiquitous, available to everybody...

DR: Right - its not the possession of individual governments or companies. Although people are very familiar with accessing the Web through computers and now mobile phones, Web technologies have a much broader applicability. If we want to have different devices networked and we want to be able to allow people to build applications around them, Web technologies will help to realize that.

SITA: How about security issues?

DR: We're still in early days. For example, banks have tended to decide it's cheaper to fix problems as they occur rather than provide improved security. That's slowly changing. The fact that we expect the user to remember a whole range of user names and passwords is not very practical. Security on the Web today is not very usable. This allows abuse, unfortunately.

Lawyers might say that if people want to access a service, you show them a legal disclaimer and everything's ok. But that doesn't really work, because ordinary people, when they see security disclaimers, do not necessarily understand them. There's a lot of potential for improvements in security and trust - maybe by using delegation models. Rather than expecting you as an individual to know about a particular service or website, maybe you can 'call a friend' or trusted authority.

Privacy is another key issue. I'm involved with a number of companies researching privacy and identity management. As you interact with the Web, a vast amount of information is being collected about you. We cannot put the clock back, but we need to empower people to control their lives. This is another big challenge for the Web. Trust, privacy and identity management are big issues.

SITA: What progress is being made with the Semantic Web?

DR: Tim Berners-Lee's first proposal for the Web used links between pages and sites that were labelled, whereas conventional links were all the same. So even then he had the idea of attaching semantics to the links. The Semantic Web is essentially about creating a Web of machine interpretable data that can be used to support systems and services.

Today corporate IT is centred around databases, but the data within the database can be likened to an island surrounded by sea. You can send a ship over to get some data and take it to another island, but they're not really integrated. We'd like to be able to connect all these things up. So the next problem is that the data within these databases is not very well described - the semantics of it is buried in code. There is documentation about the data models but it's buried somewhere else and most likely out of date.

To allow computers to know that this is an email record and that it is someone's address, you need good machine readable descriptions. Text, diagrams and pictures are things that computers have difficulty understanding. Google has shown that statistical techniques for text analysis work to a certain extent, but the kind of links you get back show the limitations to that approach. If you can describe the meaning of something in a way that a machine can understand... It's the difference between having an isolated computer to one that is networked with the world. There's a lot of activity going on in this area, with a whole slew of specifications coming through.

SITA: So if the air transport industry wants people to be able to pass from kerb to aircraft without breaking stride*, as was suggested 10 years ago, the Semantic Web is integral to that.

DR: In the long run, yes. The Semantic Web will be very important for this. Today, if you search for flights, by and large you don't get a lot of useful stuff back. If there were standards around services, then search engines could work with different industry statements to define these application programming interfaces. Today you have to go to different websites for the different components of a trip, whereas if the site could come back to you with possibilities that suit you personally at that moment very quickly... It's about understanding the intent of the search. Doing the work on behalf of the user so that users don't have to spend a lot of time on different websites. There's huge potential there.

SITA: So with a context-based search, it becomes easier and faster and more reliable to find and book a flight from a mobile, for example?

DR: Yes, but it also brings us back to the privacy issue. You might be willing to disclose information about yourself to a company in exchange for information, providing they don't hang on to the information. So we're back to the old adage about the quality of information you get back being impacted by the quality of information you give them. If the company knows your frequent flyer details, where you are right now etc, it can do a lot more.

At the same time, you don't want to give that information away too readily. So this is where privacy comes in. It's about accepting different levels of risk. You want to know whether the sites that are collecting information are trustworthy... Hence the importance of delegation models - at the moment people are trying to make judgments with incomplete information.

SITA: In summary, we're still at a fairly young stage in the evolution of the Web?

DR: We're still learning about and understanding what we have here. It's all about breaking down silos. People make investments in technologies and they have to use them for many years. But the Web allows people to put a layer in front of that and make information much more useful, more accessible. We're still only beginning to realize the full potential of the World Wide Web.

* "What do customers want? They want to be treated like royalty. They want to be able to get on the plane without breaking stride." Nobel Prize winner Dr Arno Penzias speaking at SITA's 'Celebrating the Future' 50th anniversary conference, 30 June 1999. He was talking about the idea of being able to arrive at the airport terminal and walk straight through to your flight, with radio networks able to check you in and direct you to the correct gate for departure without the need to stop, queue or interface with multiple systems.