The Portable Data Base     Jeremiad

Breaking the 3D Logjam

AutoCAD Version 2.6, which shipped in April of 1987, extended the pure extrusion “3D Level 1” of Release 2.1 by adding two new entities, 3DLINE and a planar, three or four sided 3DFACE which could have any orientation in space. These were jammed in as a stopgap so that more interesting models could be created and rendered in conjunction with AutoShade. After Version 2.6, we were working in parallel on two projects. “White Album” (Release 9) centred around the pulldown menu and dialogue box features I had implemented in “Urgent Fury”, my 1986–1987 “annual week of rest” project (this tradition is described on page [Ref]), along with the Portable Database code (see page [Ref]). The other project, called “2.7”, was to be a fundamental redesign of the guts of AutoCAD to turn it into a general 3D modeler.

The 2.7 project, as originally constituted, was making very little progress, since attempting to re-architect the internals of a product as complicated as AutoCAD while preserving complete compatibility is an extremely difficult task, especially when it involves ripping up the product in such a manner that nothing works for a long period of time. By the time 2.6 shipped, the 2.7 project was going nowhere and everybody was getting discouraged.

In March 1987 Scott Heath suggested that we abandon the bottom-up rewrite approach and instead add 3D to AutoCAD in an incremental manner. This made a lot of sense—the design he was proposing seemed eminently doable, and we'd be able to work on the release just as we'd developed previous updates to AutoCAD.

I decided that breaking the 3D logjam would require not just a good technical design, but also punching through the discouragement that was beginning to persuade people “we just can't do 3D at Autodesk.” So, indulging my flair for the dramatic, I decided to call a meeting, on a Sunday, at my house—in the same living room where Autodesk was founded—to figure out how to make AutoCAD do 3D, and how to get it done quickly. To reinforce the importance to the company, Al Green attended the meeting.

This memo was taken, by some, as “undermining the technical management,” which kind of surprised me, since the technical management was getting nowhere in 3D and I, as chairman of the company, knew that if we didn't have 3D soon we'd be out of the game. Scott had made a concrete proposal for how to build a 3D AutoCAD, and my reaction was to let him try to pull it off. This was the genesis of AutoCAD Release 10.


To: Distribution
From: John Walker
Subject: Programmers' Meeting
Date: March 5, 1987

Autodesk has lost the technical leadership in the industry that it enjoyed for the four years in which AutoCAD became established as the worldwide industry leader for CAD.

The reasons for this loss are many. Seeking to assign blame or to revisit decisions made in the past with inadequate information squanders precious resources we should be spending to regain the high ground and continue our leadership of the industry.

We must announce a full 3D version of AutoCAD at COMDEX Fall 1987, and ship it before December 15, 1987.[Footnote] If we fail to do this, we will forfeit our industry leadership, market share, stock valuation, dealer loyalty, and preeminent position as the application base for the CAD industry.[Footnote]

I have invited you to this meeting because I believe that you are one of the cadre of people who can make this happen. Programmers created Autodesk because we believed that our efforts could create products and thereby create a company much better than the companies we worked for in the past. Now programmers must come to the aid of Autodesk in the time of its greatest technical challenge.

Scott Heath has made specific recommendations as to how we can implement 3D within AutoCAD in an expeditious fashion. Tony Monaco[Footnote] has said that nothing is as important as delivering a full 3D system as soon as possible. The management and directors of Autodesk are unanimous in supporting Scott's initiative.

At 14:00 on Sunday, March 8, 1987, at my house (16 Saint Jude, Mill Valley), I will convene a meeting of those us who can help Scott to achieve this goal he has set for himself and our company. One thousand eight hundred and sixty three days before this meeting, sixteen programmers met in the same room and created Autodesk. This Sunday, another small group can meet there and rescue it.

I am not inviting you to enter into an easy task—it will be the greatest challenge we've undertaken and I expect it to be for most of us the most stressful and difficult task we've yet encountered.

I invite you to partake of a challenge that few people ever have a chance to undertake—a test of our individual and collective mettle which, if met, will yield not only material rewards commensurate to the efforts expended, but also that inner confidence that can come only from looking down into the black gulf of failure, then jumping it.

Please let me know as soon as possible if you're in. No opprobrium will be attached to passing this up. Parking is catastrophic, so I'll coordinate car pools up the mountain. Pepsi will be supplied; pizza will be ordered as required.[Footnote] Expect to leave this meeting (maybe 48 hours later) with specific task assignments which constitute the most challenging tasks of your career.[Footnote] Expect to fly back from Las Vegas this November knowing that your efforts saved Autodesk.[Footnote]

Distribution: Dan Drake, Bob Elman, Al Green, Ben Halpern, Patricia Harris, Scott Heath, Jack Heller, George Krucik, Duff Kurland, Mauri Laitinen, Greg Lutz, Eric Lyons, Tom Mahood, Steve Mighetto, Lars Moureau, Vladimir Pochop, Hal Royaltey, Kern Sibbald, Walt Spevak, Robert Wenig.

The Portable Data Base     Jeremiad