copyright notice
accesses since April 8, 1996

the State-of-the-Art in Client Server Browsing and Navigation

Hal Berghel The University of Arkansas

[figure 1]


The term cyberspace was coined by science fiction author William Gibson in his 1984 book, Neuromancer [5]. In that work cyberspace was associated with negative images: urban decay, avarice, the use of information for control. However, cyberspace has now come to take on a more positive connotation - a mass noun for progressive networking technology. Cyberspace denotes a giant information cloud unified through today's digital networks. It is a doubly-Einsteinian universe of information. Not only is it ever-expanding, finite and unbounded, the first Einsteinian twist, but it is also relativistic. What cyberspace is depends upon one's point of view and the tools that one has to experience it: to a cybernaut with an ancient Gopher client, cyberspace may appear dull and lifeless where for Webbers armed with MPEG viewers and WAV players the experience can be a multimedia awakening.

In any event, when cyberspace comes to life it does so through virtual windows on workstations supported with client-server technology. The genesis of this virtual world was a technology that appears modest by modern standards: Telnet. Along with Email and ftp, Telnet was an early Internet killer application. What made Telnet distinctive was the fact that Telnet gave us inter-connectivity, the sine qua non of cyberspace. Telnet was not just a remote login procedure in a timesharing system. It was an inter-connectivity tool for every client and every server (in principle at least) on the Internet. The viability of this one general-purpose, pervasive program paved the way for all client- server technology to follow. It was an early milestone in the evolution of cyberspace.

What followed is miraculous even by modern computer industry standards. From 1983 when Arpanet went "public" to the present, the Internet has grown to 25 million users on 2 million computer systems using 10 thousand subnets [6]. Monthly growth rates are estimated to be in the double digits. The usage statistics depicted in Figure 1 are typical for modern network information centers.

The Internet Toolbox

Internet client-server resources are difficult to classify because they are motivated by very different interests. We'll illustrate this with two examples.

First, consider the case of Internet "white pages." These resources (Whois, X.500, Knowbot Information Systems (KIS), Netfind, etc.) are directory services for the Internet itself. However, even within that small group there are large differences. Whois is a centralized directory based upon the Unix Finger program. X.500 is a decentralized directory. KIS isn't really a directory service at all, but is a directory service locator. That is, KIS uses existing directories to find Internet users. Netfind is yet an entirely different animal. It doesn't limit its domain to network directory facilities like Whois and X.500. It will scour the net for Internet users on individual servers.

Second, let's consider the case of Emailers. There are scores of Email systems with widely varying capabilities and vastly different levels of sophistication. Some, like Elm, betray a 1970's human factors orientation. Others, like cc:mail and LaMail are modeled after today's word processors. Some mix the old with the new. Pine, for example, has an interface that is clumsy and stiff by modern standards. But it allows the Emailer to attach multimedia files to the message - forward thinking even for the 1990's. Like the net directory services, they span the spectrum of that which is possible and desirable.

Figure 1. The usage statistics for modern network information centers.

So taxonomies of Internet resources should be considered in perpetual motion. Internet resources for the future will be a pot pourri of all things imaginable. That's one of the reasons why people refer to the Internet as anarchy that works.

We offer the following categorization (see Figure 2) as one of many perspectives on current Internet client-server resources.

Given our concern for simplicity, our classification scheme let's things fall through the cracks. Internet Talk Radio is a new form of digital mass communication - the Internet radio station as it were. The audio tracks of the program are digitized and stored at an Internet ftp site for downloading and playback at your workstation. This facility really falls within the rubric of mass communication via cyberspace which is not covered in this article.

The same holds true for the new wave of audio/visual offerings. These programs fall into four basic groups: teleconferencing software, whiteboard software that supports interactive document creation, audio/video annotation software that adds sight and sound to basic text/graphics transmissions, and algorithmic animation and visualization that is kind of an upscale Telnet. While presently LAN creatures, these programs will soon be client-server offerings on the net. Their importance derives from the fact that they add human interaction to the virtual terminal.

Finally, there is MUD? Multi-User Dungeons were originally network game programs. Now they have interactive educational programs bearing the MUD look-and-feel. The MUD region of cyberspace is a nebulous one.

Surfing on the Web

Besides the transmission volume, Figure 1 also revealed a significant change in environment preference over the past few years. Increasingly, the World Wide Web, or simply the Web, is becoming the surfers' vehicle of choice for trips through cyberspace. Since this trend is likely to continue for the next few years, we will focus on that aspect of the cyberspace experience. To be more specific, we will focus on only navigation and browsing utilities within that domain.

While Figure 2 reveals some variations between client-server programs, it is not complete. As any cutting-edge Web surfer will confirm, within each category there are products with significant differences. Webbers will confirm that such clients differ significantly between each other in terms of functionality. Clients in one family will have a very different look-and-feel from those in another, even when written for the same environment. There is as much variety within the family of Mosaics as there is between families.

The point is that even when one looks only at one environment, there is little orthodoxy when it comes to software design and engineering. The closest thing to a standard that we have is Mosaic. Even there the variation within products is noteworthy.

To illustrate, consider the functionality of a sample of current Web navigator/browsers (Figure 3). While the general structure of the display page is predictable due to the html compliance required by Web, major differences remain. In the Netscape Mosaic, for example, the search engine is integrated with the navigator/browser - a non-trivial undertaking and well ahead of its peers in this area. Win Tapestry supports a hierarchical hotlist folder organization, unmatched by the other products. Since hotlists and bookmarks don't scale well, this is an also an important feature for the serious user as is the ability to edit, import and export hotlists and bookmark files to foreign products. Web Explorer on the other hand has a well-developed cyberlog or itinerary history, and like Air Mosaic has a full-screen or kiosk option that is important in presentations to minimize the distractions produced by the workstations' background desktop. In short, all have advantages and shortcomings when it comes to features which serious users would consider to be important. These are obvious from even an incomplete overview such as that in Figure 3.


Predictions must always be taken cum grano salis, and those that follow are no exception. However, we think that by observing cyber-surfing behavior and continuously comparing client-server navigation and browsing aids to peer software resources in other areas, particularly in the area of business applications, one can see some fairly clear patterns emerge. We suggest that client-server navigation and browsing utilities of the future will be significantly different that those of today in the following areas:


We are now at the base of the exponential growth curve of cyberspace evolution. As commerce and industry gets connected, the range of services will escalate to bring previously unimagined resources to the computer literate. Cybersurfing will be a passing fad as the novelty wears off and more sophisticated information services are offered. Cybersurfing will become the cable surfing of the 21st century - something that we do when we're too tired or bored to do anything productive.

For the next few years, the greater interest will be with the hardware and software innovations. This is true with any technology in its infancy. As cyberspace technology matures, the interest will gradually turn away from the technology and toward the content of the information conveyed. As a society, we are in the fascination phase of this technology - much as our grandparents were with their cumbersome and unsophisticated crystal radio sets. Quickly, we shall come to expect perfection from our cyberspace products and will balk at the prospect of perusing cybermedia that has not been customized. By the end of the decade, as the bandwidth and efficiency of the nets increase, cyberspace will support a full range of interactive, participatory, three- dimensional, omni-sensory, virtual experiences. By mid-century it may be exceedingly difficult to distinguish between the virtual experience in cyberspace and its veridical counterpart.

Since Cyberspace is in its infancy and we don't yet depend upon it for our survival, it can be still be enjoyed. Like a good cruise, much of the fun is just going somewhere. Toward that end, you might enjoy leaving from my dock: url=

Acknowledgements: I wish to thank Ed Deaton and Jim Hightower for inviting this keynote address. I also express appreciation to PC AI Magazine for allowing me to include some earlier material of mine [4].


[1] Berleant, D. and H. Berghel, "Customizing Information: Part1-Getting we need when we want it", IEEE Computer, 27:9, pp. 96-98 (1994).

[2] Berleant, D. and H. Berghel, "Customizing Information: Part 2 - How successful are we so far?", IEEE Computer, 27:10, pp. 76-78 (1994).

[3] Berghel, H. and D. Berleant, "The Challenge of Customizing Cybermedia", Heuristics: The Journal of Knowledge Engineering, 7:2, pp. 33-43, (1995).

[4] Berghel, Hal, "Cyberspace Navigation". PC AI Magazine, 8:5, September/October, pp. 38-41 (1994)

[5] Gibson, William, Neuromancer. Ace Books, New York (1984).

[6] Gilster, Paul, The Internet Navigator. Wiley, New York (1993).