Andrew Treloar [HREF1], School of Computing and Mathematics, Deakin University, Rusden Campus, 662 Blackburn Road, Clayton, 3168, Australia. Andrew.Treloar@deakin.edu.au
The Web is a relatively new technology, having only been available since 1993. At the same time, it sits on top of other technologies that have been around for much longer: GUI APIs, operating systems, computer and networking hardware. The development of each of these technologies follows predictable curves as it moves from innovation through to maturity. The adoption rate by consumers also follows predictable curves and tracks the technology itself. This paper examines the position of the Web browser in this process of development and considers some of the future implications for the browser and by extension the Web itself.
Despite its relative youth, the Web has rapidly become a rich and complex set of technologies. Other papers at this conference will discuss different aspects of this technology soup, in many cases focussing on leading/bleeding edge developments. The aim of this paper is much more modest - to look at browsers (but to use browsers as a lens to reveal something deeper and more interesting).
Browser technology is nowadays fairly mundane (at least to most techno-geeks (like the author (multiple indentation levels - a dead giveaway!))). It is there, it is ubiquitous, it works, and it is becoming fairly standard. The two products with the majority of the market share grow ever more similar as they relentlessly copy each other's innovations in a leapfrog style. But browsers are still important, particularly if one wants to examine Web technology from the point of view of an average user. This is because from the user's perspective the Web is the browser. This is not to deny the importance of servers and server technology, but this technology is largely invisible to the user. Either it works or it doesn't.
Browsers are examples of software products. They are somewhat unusual products in that the two dominant browsers (Netscape/AOL's Communicator and Microsoft's Internet Explorer) are both distributed at no cost to the user. But they are products nonetheless. Product development has been examined in other (i.e. non-software) fields for decades now. Perhaps the literature of product development can teach us something about browsers. Two themes seem particularly relevant: product balance and the importance of product type.
One of the themes in the product development literature during the last decade has been the importance of centering the development around the needs of the end-user. Donald Norman (Norman, 1998) argues that human-centred products need to be balanced (like a stool - see figure 1) on the three legs of marketing, underlying technology and user experience. Each of these legs is critical and if one dominates then the stool becomes unbalanced and the product will probably fail. The Xerox Star is an example of a computer product that failed because of deficiencies in marketing and technology, although the user experience was superb.
|
|
Different aspects of the product stool are important at different stages in the life-cycle of a product. Early adopters are interested in the technology and what it can do. Marketing and user experience just have to be 'good enough'. Late adopters are much fussier and require all aspects to be in balance.
Classical economics distinguishes between substitutable and non-substitutable goods.
A good is substitutable if an alternative can be found by the user that meets their needs equally well and if the cost to switch is low or zero. Most traditional products fall into this category: cars, food, clothing, consumer electronics. Note that much advertising is devoted both to suggesting that such goods are not substitutable (Coke is not the same as Pepsi) and yet paradoxically urging users to substitute their consumption and switch their allegiance. With such traditional products, a company does not need a dominant market share. Indeed, there may be a large diversity of such products in a given market segment.
A non-substitutable good cannot easily be switched. Norman (Norman, 1998) points out that infrastructure products (like network switches or operating systems) are inherently non-substitutable. To switch from one to another carries a large cost and consumers are therefore reluctant to switch unless they have to. Infrastructure goods benefit from what economists call network externalities (Agre, 1995). They naturally tend to the dominance of one standard as the benefits to the adopters of that standard increase with its relative proportion of market share. Examples include OSI vs TCP/IP, Beta vs VHS, and MacOS vs Windows. Regardless of its deficiencies, the more people who use Windows, the more software gets written for Windows and the more attractive it becomes for new users. A positive feedback cycle starts and naturally reinforces itself.
Browsers are therefore largely substitutable products (the overwhelming majority of Web sites can be viewed by both Communicator and Internet Explorer) sitting on top of non-substitutable operating systems. How can we understand the life-cycle of browsers to date and their likely future?
Rogers classifies stages in the technology life-cycle by the relative percentage of customers who adopt it at each stage (Rogers, 1995). Early on are the innovators and early adopters (who are concerned with the underlying technology and its performance). Then come in succession the early majority pragmatists, the late majority conservatives and lastly the laggards (all of whom are more interested in solutions and convenience). Moore depicts the transition between the early adopters and early majority pragmatists as a chasm that many high-technology companies never successfully cross (see figure 2) (Moore, 1991)
|
|
Christensen prefers to look at the phenomenon of technology take-up from the perspective of the level of performance required by average users (those in the early and late majority categories in figure 2) (Christensen, 1997). He argues that once a technology product meets customers' basic needs they regard it as 'good enough' and no longer care about the underlying technology (see figure 3).
|
|
Norman argues that these two curves are in fact telling the same story (Norman, 1998). It is when the technology becomes good enough that the chasm between early adopters and late adopters is crossed (see figure 4).
|
|
So how do we get from left to right: from technology-driven to customer-driven products? Through innovation, which has its own distinctive dynamics.
The process of moving from left to right on figure 4 requires innovation. Utterback distinguishes between two types of innovation: product innovation and process innovation (Utterback, 1994).
Product innovation involves coming up with something new (or sufficiently different). Successful startup companies begin with product innovation; they either come up with a significantly better products or a new product altogether (like the first Web browser). Small nimble companies find it much easier to do this in part because they are willing to take risks (and because they have less to lose). Organisations at this stage are often characterised by what has been called the Ôhandcraft syndromeÕ: everything is done by hand with few systems in place to automate things (in part because everything is new and being developed Ôon the flyÕ). Many early Web-sites were characterised by such systems.
Rogers (Rogers, 1994) points out that sometimes it is necessary for multiple innovations to be adopted at the same time. As an example, it is not possible to introduce cars that use hydrogen as a fuel without providing a way to refuel such cars and an infrastructure to support such refuelling. Hahn and Schoch (Hahn & Schoch, 1997) call this an innovation cluster - adopters of an innovation do not have to adopt all members of the cluster but if they adopt one it is more likely that they will adopt others.
Process innovation involves changing the way the company does things to do them better (but without necessarily changing the product). Once a product is established the company needs to transition to process innovation to bring the price down and raise the quality. The initiative by Netscape to set up mozilla.org [HREF2] as a way of gaining assistance from the open source development community in contributing to the next version of its browser can be seen as an example of process innovation. Making this transition from product to process innovation is critical to moving from a high technology product to a consumer commodity product. Most commercial web-sites have transitioned (or are transitioning) from craft-cottage page production to some more highly automated, perhaps database-driven, production system.
The process shown in figure 2 is one of the diffusion of an innovative technology over time. Huff and McNaughton (Huff and McNaughton, 1991) argue that innovation diffusion (such as the takeup of use of a web browser) is best understood in terms of five major components:
This starts with an individual finding out about an innovation, then forming an attitude to it, then making a decision (adoption or rejection).
Innovativeness is a characteristic of the adopter. The categories are those listed in figure 2 above. Dickerson and Gentry Dickerson and Gentry, 1983) summarise a number of earlier studies to argue that early adopters (relative to late adopters) tend to be younger, better educated, have higher incomes and work in higher status occupations.
This generally follows the classic S-curve which can be derived from figure 2 by changing the Y-axis to cumulative percentage of users and redrawing.
These include relative advantage, compatibility, complexity, trialability, and observability.
This is the channel by which users find about innovations.
So how does all this help us understand developments in the browser? In considering where current browser technology fits within these curves, it is necessary to remember that browsers are just the top level of a technology stack that starts at the lowest level with the hardware layer, then moves to the operating system layer, then adds the network layer and finally the application layer. Each of these levels may be at different stages in the technology life cycle.
The majority of personal computers today (with the exception of high-end specialist machines and notebooks) use commodity hardware for their motherboards, processors, secondary storage, memory and displays. Every segment of the market contains multiple suppliers who are all actively innovating and competing on cost, convenience and reliability. MooreÕs law ensures that performance/dollar continues to double every 12-18 months (depending on the type of hardware). This performance curve is likely to operate into the near future (the next five years at least). This means that each year either the performance available for a particular amount increases or the cost to buy a fixed amount of performance decreases. Unfortunately, a variety of factors mean that usersÕ expectations rise at about the same rate as available performance, meaning that the computer one wants always costs about the same amount each year (although it keeps improving in performance).
The performance of the various hardware segments (CPU, memory, secondary storage, display) passed the Ôgood enoughÕ point some years ago, and most consumers now donÕt particularly care about the technical details of their system. They just want something that works. Computer games players are an exception here. Computer games are at the leading edge in the demands they place on computer hardware (particularly the graphics sub-system) and the players (and producers) of such games are continually demanding increasing graphics performance. The shift towards commodity hardware is sufficiently far advanced that a number of companies in the U.S. have announced plans to provide a free computer provided that consumers sign up to use a particular Internet Service Provider (ISP) (Sullivan, 1999) or commit to divulging personal information and using a browser with embedded advertising (MSNBC, 1999).
The development in operating systems has gone hand in hand with advances in hardware (particularly CPU speed and secondary storage). Faster CPUs have enabled a greater percentage of the CPU to be allocated to providing a better user experience (and adding unnecessary bells and whistles) and larger hard disks have supported more bloated (and more functional) operating system and application code. The Mac OS passed the 'good enough' point in the late 1980's and Windows passed it with the release of Windows 95. MicrosoftÕs dominance of the personal computer OS market means that the operating system is also now a consumer commodity. Interestingly, Linux is still in the high technology space (refer figure 3) but is rapidly transitioning into a consumer commodity via the activities of companies like Red Hat and Debian.
The network space is one where the last two decades have seen a transition from one non-substitutable technology to another. This is, of course, the transition from ISO/OSI to TCP/IP. The transition occurred for at least three reasons. TCP/IP is a superior technology in many respects, it is more open, and the ascendancy of the Internet (based on TCP/IP) drove the network standards demanded by consumers. Most such consumer users of personal computers only developed a need for access to wide area network connectivity after the transition had taken place and have therefore inherited a mature networking technology. TCP/IP is now built into all major personal computer OSs (Windows9X, MacOS and Linux) as the default WAN protocol. As with the OS, TCP/IP access is expected by users to be there and to work. Network hardware has largely standardised on 10-BaseT for LAN cabling, with fibre optics usually used for organisation or campus backbones.
The browser (as evidenced in Netscape Communicator and Microsoft's Internet Explorer) crossed the chasm on both the graphs shown in figure 4 at about release 3.0.
As far as product performance goes, version 3.x browser technology is clearly 'good enough'. In fact, as recently as last year large organisations like Telstra still had version 3.x as part of their standard operating environment. It satisfies the basic needs for browsing standard HTML and supports additional features like tables and Javascript.
As far as the number of users are concerned, it was at about version 3.x that new users of Web browsers (and by extension the Web) moved from being early adopters to early majority pragmatists. The further transition to late majority pragmatists is occurring now, with URLs routinely placed as part of advertising campaigns, listed on TV show credits and provided as part of radio broadcasts. Government departments are rapidly becoming Web-enabled and the Web as a whole is becoming part of popular culture. It is no longer the specialist niche phenomenon we celebrated at AusWeb95.
So what does all this mean? Norman's model predicts that once a technology is mature, customers want solutions and convenience. The user experience is what dominates. This suggests a number of predictions for the future development of the browser.
The first is that late adopter users will not really care which browser they use. They will assume it does what it needs to do: navigates the Web, renders HTML (or its successor), supports Javascript (or ECMAScript) and Java. All of this will be a given, in the same way that we take it for granted that a car can transport us safely from point A to B. The challenge is therefore for browser developers to focus on the user experience. One (regarded by most commentators as failed) attempt to do this is Microsoft's Active Desktop introduced with IE 4.0. A more successful (albeit minor) example is the form auto-fill feature in EI 4.5. The prediction here is for an increase in developments in the Web browsing user experience, some of which may change the document-centred browser model we have at present.
What then of the role of innovative Web technologies like CSS, SMIL and XML? Most users are completely uninterested in such things, although they may be interested in the new things they permit. The distinction between a Web document rendered in HTML 4.0 and the same document enhanced with CSS falls into the realm of excess quality (shown on figure 3) for most late adopter users. Most such users are uninterested in the improvement in quality; single column, single type is 'good enough'. The prediction is that implementation of these technologies will be technologist-driven rather than user-driven, and focussed on process innovation. Unless such technologies can deliver a significantly improved user-experience, users will not clamour for them. Examples of such an enhanced user-experience made possible by XML might include richer annotation features using X-pointers and improved searching using RDF metadata. Other features that XML can make possible such as automated generation of tables of contents and improved link maintenance will be largely invisible to the user.
A number of commentators, including Jakob Nielsen, have been arguing recently that moving all information systems to the Web and using the browser as the universal client is a deficient model. He argues (Nielsen 1998) that this will deliver a flawed user experience and that providing rich information access is better done with a dedicated (and Internet-enabled) client. The consequence of this is that the browser will retain a useful role as the universal navigation gateway to online information, but not as the universal client to all information found online. Other clients (such as ICQ, Hotline and Carracho) will continue to be developed for particular requirements.
The model of technology life-cycles outlined in this paper enables us to better understand where the browser has come from and where it is going. Understanding the implications of the shift from high technology product to consumer commodity will be critical to planning for future changes in the browser, and by extension in the Web itself.
The author wishes to acknowledge the very useful comments (and the consequent improvements in this paper!) made by the two anonymous referees.
Agre, P. E. (1995). Designing genres for new media: Social, economic, and political contexts [HREF3]. The Network Observer, 2(11).
Christensen, C. M. (1997). The innovator's dilemma: When new technologies cause great firms to fail. Boston: Harvard Business School Press.
Hahn, K L, and Schoch, N A. (1997). Applying diffusion theory to electronic publishing: A conceptual framework for examining issues and outcomes[HREF4]. Proceedings of the ASIS Annual Meeting 1997, 34, pp. 5-13.
Dickerson, M. D. and Gentry, J. W. (1983). Characteristics of Adopters and Non-Adopters of Home Computers. Journal of Consumer Research, 10 (September). pp. 225 - 235.
Huff, S. L.and McNaughton, J. (1991). Diffusion of an information technology innovation. Business Quarterly, 56(1), pp. 25 - 31.
Moore, G. A. (1991). Crossing the chasm: Marketing and selling high-tech goods to mainstream customers. New York: HarperBusiness.
MSNBC (1999). Free PC offer gets plenty of takers [HREF5]. MSNBC, February 8, 1999.
Nielsen, J. (1998). Does Internet = Web? [HREF6] Alertbox, September 20.
Norman, D. (1998). The invisible computer: why good products can fail, the personal computer is so complex, and information appliances are the solution. Cambridge, MA: MIT Press.
Rogers, E. M. (1995). Diffusion of innovations. Fourth edition. New York: The Free Press.
Sullivan, B. (1999). Free Net access gains steam [HREF7]. MSNBC, Feb. 8, 1999.
Utterback, J. M. (1994). Mastering the dynamics of innovation. Boston: Harvard Business School Press.
Andrew Treloar, © 1999. The author assigns to Southern Cross University and other educational and non-profit institutions a non-exclusive licence to use this document for personal use and in courses of instruction provided that the article is used in full and this copyright statement is reproduced. The author also grants a non-exclusive licence to Southern Cross University to publish this document in full on the World Wide Web and on CD-ROM and in printed form with the conference papers and for the document to be published on mirrors on the World Wide Web.
