Glasnost

Throughout 1987 I had been becoming increasingly convinced that Autodesk's long-standing policy of tight security regarding new product releases was growing counterproductive. My experience with Univac mainframes, where there was essentially full disclosure to the customers of Univac's software development plans, persuaded me that such an environment was far more conducive to developing products that met the customers' actual needs.

The question always was, “can we afford the financial risk of disclosure followed by a late shipment, and the competitive risk of laying out our development plans for all to see?”. In this memo, I argue that we can. Pursuant to this proposal, for the first time we briefed our dealers and developers on the features to be announced in AutoCAD Release 10 (Abbey Road), well in advance of its official unveiling at AutoCAD Expo in May of 1988.

To: Al Green, Dan Drake, Malcolm Davies, Chris Record, Eric Lyons
From: John Walker
Date: 20th February 1988

Subject: Гла́сность

For some time I've felt that our policy of high security with regard to the features to be included in future releases of AutoCAD and general information concerning their release schedule, was growing increasingly outmoded in light of our dominance of the market and our desire to maintain a close cooperative relationship with our resellers and their customers. In this memorandum, I'd like to recap the reasons that caused us to originally adopt our policy of secrecy and then examine the changes in the competitive environment which have, in my opinion, caused this policy to become not only unnecessary, but actually damaging to our competitive position.

In this document, I will present the case for changing our policy to one of essentially complete openness and full disclosure regarding our plans for the future development of AutoCAD; moving toward a cooperative relationship with our user community more like that between a mainframe computer vendor and its user group, or MacNeal-Schwendler and the NASTRAN user group. I will present this case in the manner in which I communicate best: as an advocate. The fact that I present a strong argument here should not be taken as an indication that I harbour no doubts about the possibility of this policy backfiring, or that I am unaware of many equally strong arguments for continuing our current policy. My purpose in presenting this case is to make us reconsider what we are trying to accomplish with our current policy and then rationally decide what changes may be in order, instead of blindly maintaining the status quo.

The original rationale for maintaining security when a new release of AutoCAD impended was simple survival of the company. Since we never offered any formal inventory protection to our resellers, and because each release of AutoCAD contained a large collection of additional features, news of an upcoming release would cause dealers to defer orders until after the release became available. On several occasions when news of a forthcoming release leaked out, we saw our sales drop to half their previous level for the one or two months before shipment of the new version.

Since the company was so thinly capitalised, premature announcement of a new product followed by a shipment delay, for whatever reason, could place the company at risk of bankruptcy. Consequently, the prudent course was to maintain tight security about the content and schedule of upcoming releases. We even went so far as to try to eliminate notification clauses from OEM agreements we negotiated. Obviously, we had no desire to create ill-will by loading up resellers with soon-to-be obsolete product, but neither could we run the risk of sinking our company.

After the 1985 initial public offering, we had enough money that order deferral couldn't bankrupt us, but another imperative asserted itself—the need to report quarterly increases in sales and profits. Since we had experienced one- to two-month 50% sales fall-offs, followed by compensating sales bulges after shipment, it became essential that if one of these was to occur, it be totally contained within a fiscal quarter. Otherwise, the Wall Street isomorph of wind shear could buffet Autodesk's standing in the financial community.

So, our policy was total secrecy about what we were working on, and “no comment” regarding shipment dates. This policy was the product of the environment in which we operated at the time, and served us well. But I believe that it may have now become outdated, and may actually be working to our detriment.

At the time we adopted the policy of secrecy, we were a small, virtually unknown company attempting to establish itself in a market which was largely unrecognised by mainstream microcomputer analysts. Today, we command over 60% market share in a rapidly-growing market which is viewed as one of the primary applications for the high end of the desktop computer market. Autodesk's installed customer base dwarfs any competitor, our distribution network essentially controls the route of engineering software from producer to customer, and we have encouraged the development of a burgeoning market in third party software, hardware drivers, education, books, and consulting. Our goal was to become an industry standard, and I believe that we have achieved that goal. Now, I'd argue that it's time to start acting like an industry standard rather than a marginal producer, and that so doing can reinforce our position as the standard.

But enough generalities: let's look at AutoCAD Abbey Road, and examine the consequences of the two approaches on our market. First, it's worth noting that Autodesk took a totally unprecedented step when we announced, concurrent with the launch of Release 9, that Release 10 would contain full three dimensional capability and would be shipped in the first half of 1988. Never before had we so explicitly called our shots regarding the feature content of a release and its shipment date. The reasons we did this were obvious—we'd been assailed by every competitor (even those without a product, or those with somebody's else) for “not having 3D”, and after a long hiatus during which we had no marketing presence whatsoever, we felt we had to respond with an explicit statement of when we would “have 3D”.

This step was actually part of a series of liberalisations of our product secrecy which had occurred over the preceding two years. For releases 2.5, 2.6, and 9 we had formally briefed third party software vendors prior to the announcement, and given them pre-release copies of the product to adapt their products to. We did this even though many third party developers are also dealers, and hence could adjust their own ordering schedule based on the knowledge of the new release (yes, we never said when, but when you call in the developers 60 days before AutoCAD Expo, it isn't very hard to guess the launch date). In several OEM contracts we agreed to disclose new release dates in advance. Prior to the release of 2.6, we described the product in advance and demonstrated it to industry analysts. Since the White Album announcement, we've shown Abbey Road to selected subsets of our constituencies based on our feelings for the market. We've shown several customers in-depth previews of the product, and we've used their reactions to guide our development of features for inclusion in the final product. We've briefed book publishers about the features in Abbey Road. And most recently, of course, we have held a very successful pre-release disclosure to our ADI developers, notwithstanding the fact that many of these developers also support competitive CAD products.

What, precisely, are we gaining by withholding the details of what we are doing in Abbey Road from our general user community? I believe that the expectations for Abbey Road are much lower than the reality. I don't think that anybody really expects that we'll deliver multiple on-screen views, smooth surfaces, and database handles in addition to the already announced generalisation of AutoCAD to three dimensions. The total impact of the seamless growth of AutoCAD to 3D—the product of Scott's genius—cannot be appreciated until seen, or better yet, used. If we're in a competitive environment where we've allowed our competitors to create the impression that they have 3D and we don't, then why in the world do we hide the wonder of our 3D release under a bushel? The mere announcement of the feature set for Abbey Road, which could consist of a pre-release of the manual supplement working document would, it seems to me, erect an insurmountable barrier in front of any competitor who based his sell on “AutoCAD lacks 3D”. We've already said what, 3D; we've already said when, first half. Why clam up on the details when those details will end all uncertainty about our fully supporting 3D?

Now let's step back from the details of Abbey Road to the more general question of disclosure of our future plans for AutoCAD. And it's AutoCAD I'm talking about—I see nothing to be gained and much to be lost by talking about new products in advance. But if we really believe that AutoCAD has become an industry standard, almost by definition unassailable as long as its vendor maintains a technological lead, then shouldn't we make overt the partnership between the vendor, reseller, customer, and third-party value-adder community which is the essence of a de facto standard?

In one stroke, we'd be moving from the inherently adversarial relationship of those who know with those who must guess to a partnership—a partnership forged from common interest in the future of AutoCAD: those who make it, and those who have bet their companies on it. I think it would lift a veil of implicit conflict almost as obscurant as that of the hardware lock, and with no more consequences on our financial results than our expunging that particular bleeding sore. It would also be a brassy statement of self-confidence mixed with humility: we'll tell you where we're planning to go, and we're listening to you to hear if we're solving the problems you feel are important.

It would also give third party developers a clear message on what areas were safe from Autodesk's future development, and which areas had only a limited window of opportunity before Autodesk entered them. At the technical level, knowing the design of some forthcoming features might be very helpful to a developer in the midst of his own long-term project. For example, knowing how we're doing database handles would guide those who we expect to use them. This communication would not just benefit the developers, however. At every single developer briefing, we discover some oversight in what we've done that would have made the feature much more useful, but which it's too late to change and maintain the shipping schedule. Open, two-way communication would let us remedy these shortcomings before it was too late. After all, the very best guidance on the design of a product comes from the people who are actually using it, and by developing behind a wall, we insulate ourselves from much of the interchange of ideas which could help us provide the best solutions.

From a political standpoint (using “political” in the sense I define it “the means by which groups of three or more humans interact”) this would be a golden opportunity to emerge from the marketing disarray and lack of clear strategy of the last two years with a turn that would leave our competitors without a card to play. I would envision Malcolm Davies walking onto the stage in Chicago at the opening session of AutoCAD Expo and saying something like:

“Up to now, Autodesk have been secretive about what we were doing. No more. At this show we are introducing AutoCAD Release 10, extending AutoCAD to fully general three-dimensional model creation. We are also showing AutoCAD on the Macintosh II, OS/2 Release 1, the Sun 386i, and Sun 4, providing a wide variety of hardware options for AutoCAD users who are presently limited by existing personal computer hardware and operating systems. In the sessions that follow we will describe the directions we see for the future development of Autodesk products, and invite your suggestions and comments. From here on, we'll talk openly with you about the composition and scheduling of Release 11, Release 12, and those that will follow. You, our customers, have defined the features we have included in AutoCAD ever since we started, and we seek a two-way dialogue with you.

“You have honoured us by choosing our products for your design work, and your choice has made AutoCAD the worldwide standard for computer aided design. Now it is time for all of us: Autodesk, the users, the resellers, the developers of applications, the manufacturers of hardware that supports AutoCAD, the user groups, the educational institutions that teach AutoCAD, and the authors and publishers of books about AutoCAD, to work together to continue to refine, expand, and develop this standard we have in common. We hope that by making our development plans public, we will stimulate the kind of two-way interchange with the rest of the AutoCAD community that will result in a better product which will serve us all well.”

So that's the case for going public with the future of AutoCAD, and how much like “going public” it is: full disclosure, high risk, and much greater visibility after you take the step. But we did pretty well by going public in the financial sense, and we might do just as well by taking that step in the product development arena as well.

If we end up adopting this path, we should immediately undertake the перестро́йка which will be required to roll out this policy at Expo.