E-Mail and the World Wide Web - time for an upgrade?

March 2004

When we talk about the Internet, we are often referring to the World Wide Web and E-Mail. They are undoubtedly the killer applications which catapulted the Internet into the commercial sphere, but they are both showing their age in one way or another.

The technical protocol used by Internet E-Mail was proposed by David H. Crocker from the University of Delaware back in 1982, making it one of the oldest pieces of the technical jigsaw remaining fundamentally unchanged. The popularity of E-Mail is undoubtedly rooted in its simplicity - in concept and in use, and the current popularity of SMS and Instant Messaging runs in the same vein. Ironically, the resulting ubiquity of E-Mail may prove the downfall of the underlying protocol, since the promises Bill Gates is currently making in relation to the elimination of SPAM will probably require some fundamental changes to the underlying machinery.

There is no doubt that something has to be done to reduce the amount of unsolicited E-Mail we receive every day, and there is also no doubt that Microsoft can devise an ingenious mechanism to suit. What will be more of a challenge is migrating the entire world - individuals, organizations and those managing the underlying infrastructure to this new scheme - and first we need to get the world to agree on the details.

Tim Berners-Lee put the first site live on the World Wide Web (info.cern.ch) on August 6, 1991. Since then, the Web has progressed not only in scale but in nature, and the underlying technology has been developed and adapted to serve many different purposes. Many of today's most popular web sites are not simply collections of documents but are really large-scale applications. Take eBay - a real-time, fully interactive trading platform supporting millions of items and concurrent users and connected to all manner of other related systems.

As the applications delivered over the Internet increase in complexity, it is apparent that the use of web technology as it stands is going to become a limiting factor - a weak link in the chain. Although sites such as eBay have proven what is possible, rarely do web sites come close to providing the responsive and reactive user interface provided by a well-designed desktop application running in Windows, on a Macintosh or using a Linux-based desktop. There is a subtle twist here. Linux desktops, and later incarnations of those provided on Macintosh machines, are based on a system called The X Window System. X was developed in the mid-1980s at MIT as part of their Project Athena Distributed Desktop project. In a nutshell, the X Window System allows an application running on one machine to show its interface and interact with the user at another. It works by sending information about the layout of the screen, and the actions of the user, back-and-forth across the network. This has several benefits: many users can use the application at once, and the applications do not need to be installed locally. It is, in many ways, another way to provide the same sort of architecture that sites such as eBay use, but using a set of technical protocols designed specifically for the job.

Microsoft are, not surprisingly, fully conversant with the advantages of such an approach. The next generation of Windows, known currently as Longhorn, promotes a similar architecture, albeit based around current technologies such as XML, and placing a stronger emphasis on security. If this technology progresses as Microsoft intends, this will allow us to use all manner of applications on a service basis, rather than having to install and maintain the applications locally.

As well as a technical revolution, this would bring with it significant change in the way we pay to use applications. Rather than buying a word processing package, we could be charged per minute, per document, per word, or even per feature.

Should they take place, these changes would also provide advantages within the enterprise. When embarking on a new IT software project, often we have to decide between either a desktop or a web based application. This is often a difficult decision to make, since there are compelling arguments in both directions. Usability, flexibility and responsiveness are usually attributed to the desktop application, whereas ease of management and deployment, modest hardware/network requirements and ease of remote access are all advantages provided by the web-based approach. The Longhorn approach, or even that provided by the X Window Systems, has the potential to provide the best of both worlds.

Longhorn is still some way away - it will probably be 2006 before we see the first versions of the new software. In the mean time, it is important to architect software systems in such a way as to leave the options open. In particular, software can be built with the user interface being a 'layer' which can in future be peeled-back and replaced without compromising the rest of the system. For this approach to be practical, careful attention needs to be paid to the architecture of the system, since the layer we may need to peel-back needs to be kept as thin as possible. Here once again is an example of what had been considered an outdated approach being rejuvenated and brought to the fore. Furthermore, Linux advocates will be sitting back smugly in the knowledge that their favoured platform has worked in this way all along. On the face of it, it would appear that Bill Gates has had to swallow his pride and admit they were right - although he will no doubt add a healthy sprinkling of proprietary spice to ensure that the flavour is to his liking.