Logging On


An Open Letter to Bill Gates

Network Computing

January 1, 1991

Dear Bill: We find your "information at your fingertips" vision of our collective computing future, as expressed in your keynote speech during last November's COMDEX, exciting, but fundamentally flawed.

We object to three hypotheses stated in your speech: the notion of the desirability of PC-centrism in business computing; the concept of the essential and universal nature of the graphical user interface; and the idea that Microsoft will define the standards for the new object-oriented and multimedia technologies demonstrated during your speech.

The end user has indeed been empowered with strong personal computing tools. However, the idea that those tools can or should expand and take over functions now performed in the glass house is unsound.

We believe that the war between mainframes and micros is over and that the network has won. In other words, the choice between PCs and mainframes should be left to whatever machine can do the job best. What is important is that this machine can be connected to other corporate computing resources in sophisticated ways, managed appropriately, and supported cost-effectively.

None of this is really mentioned in your speech.

The idea of network computing is that the universe has no center. Applications reside on the platforms that can best support them, and their services can be accessed by anyone on the network authorized to do so. As our feature stories in this issue make clear, businesses still need mainframes and larger-sized machines and will still use them well into the next century.

In many cases, mainframes and other larger machines still provide the most cost-effective solution to storing, manipulating and networking large data sets.

Few corporations the size of Microsoft run their businesses entirely on microprocessors. We venture that a mainframe might actually be more cost-effective for Microsoft than the several hundred PCs stacked up in your data center.

In any event, it is integration at the network level and not at the standalone PC level that has enabled companies to move applications from mainframes toward more distributed platforms. As networks become more sophisticated, reliable, secure and remotely manageable, this trend will continue. Until network integration of applications between micros, mid-sized machines and mainframes gets better, these developments will fulfill narrow niches.

Let's move on to graphical interfaces. Here again you present a case that - while visually exciting and stimulating - is far from the corporate realities that we have experienced. This is not to say that we are character-mode DOS bigots; indeed the majority of us prefer Macintoshes, Windows, X-Windows and even the flogged horse of Presentation Manager, and use them in our daily work at this magazine.

The wonder and power of these "what you see is what you get" interfaces isn't the point. If network technologies can't support them, if users can't easily gain access to networked data using them, and if the data center folks can't easily manage and configure the machines running them, then all that visual stimulation is nothing more than the '90s version of video arcade games.

What prevents Windows and Presentation Manager from becoming universal is not that people lack easy ways to access and integrate data or lack mice and graphics hardware to run them. And it is not that folks would rather type in command line arguments. Rather, it is that the resource cost for corporations to configure and manage several thousand networked workstations doesn't yet balance the business benefits of using these tools.

And, apart from the resource cost, the application interfaces are still far from where they need to be for corporations to develop mission-critical applications. You mention the next steps in developing these interfaces in your speech. But we think you still have a long way to go toward improving existing interfaces. Look how long it has taken you and others to agree on implementations of dynamic data exchange.

These are critical problems and will prevent people from moving to next-generation interfaces that involve the linking and embedded objects mentioned in your speech. They are also powerful reasons to keep folks in character mode for years to come. Look how many people are still using 1-2-3 as their principle word processor and database. Now, that's a user interface that is nearly universal.

Our final points concern the role that Microsoft will play in standards development for these new multimedia and object-oriented technologies. You mention that Microsoft is proceeding to "develop" standards. It appears that consistency with upcoming versions of Windows will define these standards. But standards aren't established simply by being proposed or developed, or even with significant resources expended on many developers' conferences. Technologies become standards through endorsement, either by a formal standards-setting agency, or - in the case of a de facto standard - through overwhelming acceptance in the marketplace.

We suggest that you learn from Apple's mistakes. Apple has enforced consistent command syntax and pull-down menus across a broad range of third-party applications, but at the expense of interoperating with non-Apple products. This concern for multi-platform interoperability is nowhere voiced. It looks, instead, as if Microsoft is planning on building a system for Intel-based PCs and letting the rest of the world fend for itself.

Finally, even assuming that interoperability between different platforms is somehow achievable sometime in the next five years, where is the bandwidth necessary for passing around all the information created by these graphical, multimedia applications?

Without the necessary bandwidth, multimedia applications must remain as standalone applications. And as such, they will remain nothing more than corporate curiosities.

But, putting on a multimedia show is one thing; bringing the vision out of the surrealistic and into the real corporate world is quite another. We don't see Microsoft addressing any of these vital networking issues - issues that, at a fundamental level, must be addressed before multimedia can ever become a reality.

We do agree with some parts of your vision. Though you've done much to make Windows more network friendly with the 3.0 version, installing and running Windows on networks should be easier. Doing so should not require four-page Workshop pieces in this magazine nor online bulletin boards with the latest patches and drivers and expert tips.


The Editors

Copyright 1991 CMP Publications, Inc. All rights reserved.