Next Up Previous Contents Index
Next: ``Autodesk Arrogance'' Up: The Autodesk File Previous: Conclusion

The Hardware Lock Strikes Back

       

I usually don't believe in premonitions. I spent Monday, January 29th, 1990, as I spend most days--sitting at the keyboard working on software development. Usually, I'm as happy as clam while doing this, but that day I felt a most unusual, unfocused mild anxiety: as if something bad were about to happen, but I didn't know what. In the early afternoon, the little flag went up on my electronic mailbox, and, reading the missive that arrived, discovered that Autodesk had decided to re-introduce the hardware lock in domestic AutoCAD Release 11. There are some decisions that are just so dumb I run out of adjectives, and this was one of them. I was also taken aback, given that I was the person who designed the original hardware lock, built the first prototype on my dining room table, ultimately, as president of Autodesk decided to introduce it in international versions of AutoCAD 2.1 and in domestic versions of 2.5, took the heat when U.S. customers arose against us (see page [Ref]), and finally decided, in November of 1986 to remove it (see page [Ref]), that nobody had consulted me or even informed me of this decision to reintroduce the lock. The announcement of the reintroduction of the lock characterised the lock as a ``deep emotional issue,'' as if to characterise those who opposed it as irrational, rather than, based on personal experience, feeling that re-locking the product would not be in the company's best interests.

This is the only instance when, after leaving the management of Autodesk, I drew a line in the sand and said, ``I am going to stop this.'' I immediately posted the following argument against the reintroduction of the hardware lock, and began to organise every resource I could summon to prevent Autodesk from jumping, once again, off the same cliff it had only four years before. I had not only written the copy for a full page advertisement addressed to Autodesk shareholders to run in The Wall Street Journal, I had obtained, from their Palo Alto offices, deadline, submission format, and price information. The next day, January 30th, Al Green reversed the decision and announced that AutoCAD Release 11 domestic would remain unlocked.

First Strike

Date: Mon, 29 Jan 90 13:56:05 PST
From: Ron McElhaney
To: Release 11 Developers
Subject: Meeting on Thursday

As you may know, the company has decided to lock ACAD both internationally and domestically. The h/w lock which will be used is the Rainbow parallel lock, with which we have had some experience, and which is in wide use by Intergraph on their MicroStation product.

The fact that this is such a deep emotional issue makes it a difficult one to discuss. Most of you will have very strong opinions concerning this decision. Much of that strong feeling is derived from the very painful ``exercise-in-futility'' which characterized the last attempt to lock AutoCAD a few years ago. In fact, many of you were here at the time, and formed your opinions first-hand. Things are different now, both in terms of the nature of the business itself and in the commercial use of the h/w lock on CAD products, but the emotional content of the issue itself is still high, undiminished by the distance of time.

I have been asked if there are any technical issues which would interfere with our ability to ship a locked product, or which would cause the release of R11 to be significantly delayed. It has been very difficult to separate out the emotional reaction to such a question and to achieve an objective assessment of the effect of this decision on the release of R11. Speaking objectively, the current (although not final) conclusion seems to be that the purely technical issues (which don't require re-working the lock challenge mechanism to make it more secure) are not significant and should not, by themselves, cause a significant delay to R11.

Unfortunately one never deals with purely technical issues in developing software. Getting R11 out on time is a matter of developing software, to be sure, but it is also a matter of deep personal committment to a goal, and a level of self-motivation and hard work which traditional companies very rarely see, but which we see at Autodesk almost all the time. What will the total impact to such a software development organization; an organization which already is working at a level which can only be described as ``hyperdrive'', and whose pace can be maintained, but only for a carefully-calculated period of time without burning everyone out.

The fact that this decision automatically evokes emotional responses means that is must be widely discussed and explained. Knowing how and why a decision was made, and then disagreeing with it, is much different than merely disagreeing with it based upon an automatic response to it. I have asked Al Green and Malcolm Davies to give a presentation to the R11 developers, for the straightforward purpose of talking about why this decision was made, and how important they feel it is to the success of the company for us to do this. Much of the impetus for this decision arose from the increasing inability of our dealers to be commercially successful representing Autodesk, and Malcolm has invited at least two of those dealers to be present at the meeting to tell their side of this very complex story.

I have asked that the meeting be held on Thursday at 1:00 in the Tech I conference room. Please give very serious thought to the issues which this decision addresses. It is important that you be prepared to discuss your concerns and objections.

Please be there.

--ron

Second Strike

Blows Against The Hardware Lock
by John Walker
Episode II: January 29, 1990

I disagree with the statement that the hardware lock is a ``deep emotional issue.'' I consider the issue of whether to lock or unlock a product to be a straightforward business decision which should be made, like any other decision, based on the company's overall goals and strategy, and from the best information at hand about all factors involved.

I would, therefore, like to ask the following largely non-technical issues related to the decision to lock the product.

1. How does adoption of the hardware lock benefit our customers?  

Autodesk has grown and prospered by always, as much as possible, placing the customer's needs foremost. By customer, I mean the user of AutoCAD, not the reseller, even though we do not sell directly. If we do not satisfy the customer, new customers will not come to Autodesk resellers, but will purchase other products instead. The hardware lock does not benefit the customer in any way of which I am aware.

2. What effect will the hardware lock have on Autodesk's sales and earnings, and on the sales and earnings of our resellers?

It is rare in business to be able to definitively answer a hypothetical question about financial results. One of the few benefits of our Dark Night Of The Soul in 1986 was that we learned the answer: None. Sales did not go up or down when we introduced the lock, and sales did not go up or down when we discontinued it. I know of nobody who predicted this result; certainly I did not. I do not believe that any material changes have occurred in the market since the last time around; in fact, the events since have moved further away from protection devices and schemes.

If the lock is reintroduced with the goal of increasing our sales and earnings and those of our dealers, and consequently improving the business viability of the AutoCAD resellers, I believe that decision to be based on demonstratedly incorrect premises. And if not with those goals in mind, then why?

3. What effect will introducing the lock domestically have on the shipment date of AutoCAD Release 11?

We are presently in the middle of one of the most furious pushes to shipment in the history of the company. Only with Stakhanovite exertion and more than a little luck do we stand a chance of meeting the current release date goal. Already, only for the second time in the company's history, we have disabled features already developed and integrated because we lack the time and manpower to debug and test them as part of the product by the release deadline. Now it is proposed that we throw another major twist into the product cycle. The statement that the lock is already part of the product is abject nonsense; anybody who lived through 1986 will recall that the logistics of acquiring, inventorying, and quality testing locks in quantities adequate for our domestic business involve problems that are not small ones. The vendors involved may be more mature than those we used in 1986, but the volume of locks we will require is also much larger. All the issues of documentation, installation problems, compatibility with hardware, and the like are identical. We introduced the lock in the development cycle of AutoCAD 2.1, which was, before Release 11, our most critical date-driven release (being scheduled near the time of the initial public offering). I think it is more than coincidental that AutoCAD 2.1 was the worst release of AutoCAD we have ever shipped, with diversion of company resources into lock-related issues a major contributor, if not the proximate cause.

4. Who are the additional customers who will buy a locked product?

Introducing the lock increases our cost of goods (perhaps doubling it, if some of the numbers I've heard bandied around are to be believed). Therefore, if margins are not to fall, additional sales must be generated (indeed, as noted in item 2 above, unless the lock is a public relations exercise or a moral crusade, this is the only reason for considering it). Therefore, what is the profile of the customer who will buy a new AutoCAD from an AutoCAD dealer if the product is locked, but would steal the product were it not? People tell me I have a fairly vivid imagination, but I cannot come up with a sketch of a sufficiently large population of customers representing that foregone revenue.

Remember that shortly after the lock is reintroduced, products will appear that circumvent it. (The suggested lock, and its implementation within AutoCAD, will be much easier to defeat than the lock of 1986). If our 1986 experience is a guide, the market price of the lock-defeating programs will be less than $100. In that environment, the question becomes this: ``Who is there who today would steal our product, but who will pay $2000+ for a legal copy of AutoCAD rather than purchase a $100 program that lets him continue to steal it?'' Legal remedies against lock-defeating programs are probably impossible and ineffective in any case.

I would suggest that those advocating the lock without pondering the full implications of this issue (which I did not appreciate until tutored by brutal experience) are victims of the same kind of naive reasoning that suggests that if one doubles taxes, tax receipts will also double. The real world is a complex web of nonlinear relations and interconnected feedback loops with delay. When you change a parameter, you're shifting incentives, not results, and you have to think out all the consequences, not just the obvious first-order ones.

5. What does the hardware lock decision say about the direction of the company?

This is not just an issue important to those who presently work for Autodesk; it is central to the overall mission and strategy of the company, its relationships with its resellers and their customers, with the marketplace and our competitors, and with the perception of Autodesk among all these constituencies.

Autodesk has always adopted a strategy of broad market share, low price, rapid enhancement, and responsiveness to the user. Almost since inception, AutoCAD has been the ``safe buy'' because so many other people use it, even if some have not yet paid. This strategy has served us well, and I am absolutely convinced that, at least in the domestic market, rampant piracy has substantially contributed to our current dominance of the market. I wish there were a way to measure the number of current legal copies of AutoCAD that replaced pirated copies. It would not startle me to discover that number to be very large.

Reintroducing the lock sends a message that Autodesk has changed its strategy. I'm concerned here with the message, not the strategy itself, and I consider irrelevant on this point comments not grounded in the 1986 experience. Regardless of the moral and intellectual merits of the arguments employed, the simple fact was that the lock was perceived as Autodesk abandoning ``the little guy'' responsible for its initial success (and I got at least as much of this from dealers, who we were trying to benefit, as from users).

This comes at a time when Autodesk is sending many other signals that seemingly herald such a change in course. We have opened regional offices. We have announced an aggressive discount program for the Fortune 500 and the government (and if you want to get onto morals, I think the concept of selling to General Motors for less than the price paid by a one-man consulting firm is as least as reprehensible as profiting by software piracy, especially when practiced by a supposedly entrepreneurial upstart company). We have concentrated development on network licenses, again aimed at the larger customers.

It is worth reflecting on the fact that despite all the hoo-rah about Fortune 500 and government sales, they're still about 15% of our business (and presumably a much smaller component of our typical reseller's business). Sending the wrong message to the people that are responsible for 85% of our revenue can be disastrous. I saw the unanticipated misperception of our intent in 1986, and I believe nothing has changed. If anything, the scars of that experience have hypersensitised our users to the issue of locks.

6. What will be the effect of the lock on the perception and initial acceptance of Release 11?

Version 2.1 was one of the most significant product introductions in the company's history. With the addition of AutoLisp, AutoCAD set itself on the course that has brought us all here today. (Full AutoLisp was deferred until 2.18, an update release, because of time-driven pressure to ship...sound familiar?) Yet the major enhancements in the product were simply lost in the furor surrounding the lock. In fact, virtually nobody paid attention to the capabilities of the product at all; it was just locks, locks, and more locks.

Now if we're pushing Release 11 because we believe both that the features it contains will increase sales and that the mere fact of a new release contributes to revenue, then can we afford to have the normal publicity attendant to that release consumed by a second hardware lock firestorm?

I urge you to think carefully about this issue. If Release 11 is so critical as to justify the efforts and compromises attendant to its push to shipment, the risk of a public relations disaster coincident with shipment is grave indeed. Having presided over the last one, I consider anybody who minimises the risk of repetition as uninformed, misguided, or naive. It may not happen, but what are the probabilities? How did you arrive at your estimates? What happens if you're wrong?

In addition, the 1986 experience demonstrated that Autodesk will cave in on locks when an uproar arises. (The argument that only a small percentage of our current users were users in 1986 is fallacious; information about the 1986 episode will be spread through all available channels once the lock surfaces. It's in The Autodesk File, after all). Consequently, there are strong incentives to:

a)
Make a big stink about the lock, because that made us remove it the last time, and,
b)
Defer buying Release 11 both to ``send-em-a-message'' and because one feels the probability is high that one will be able to buy (or steal) an unlocked version within a few months anyway, once Autodesk concedes defeat this time around.

Are these the incentives we want, or need, associated with Release 11's introduction?

Now these are just the obvious issues related to the domestic reintroduction of the hardware lock--a compilation off the top of my head based on the 1986 experience. I have not thought through the numerous, more subtle, consequences that appeared in 1986 in terms of the current proposal. I presume that in the process of arriving at the decision to reintroduce the lock, management considered all these points and more (and if not, then the decision was arrived at through a dangerously flawed process). I am therefore interested in the conclusions that were reached in regard to each of the issues I raise herein, and the background information and reasoning used to arrive at them.

If any emotion is involved in the issue of the lock, it is emotion that springs from seeing our company about to repeat the single worst experience in its entire history, with a probability of disaster I would estimate on the order of 85%. It is emotion born of seeing a company forget its past, its goals, its customers, and the principles and processes that made it so successful in the first place.

Escalation: Eyeball-to-Eyeball

Date: Tue, 30 Jan 90 15:40:29 PST
From: John Walker
To: Al Green
Subject: Letter re: hardware locks
Cc: Malcolm Davies, Dan Drake, Greg Lutz, Ron McElhaney

As a selling shareholder in the S-3 offering of June 20th, 1989, I remain personally liable until June 21, 1990 in shareholder suits based upon inadequate disclosures in the prospectus.

Should the hardware lock suggestion go into effect and cause a catastrophic drop in the stock price, shareholder suits are a virtual certainty.

To those unaware of the nature of my current day-to-day activities, and know of my personal involvement in the last hardware lock episode, documented in ``The Autodesk File'' for all to read, any claim that I had nothing to do with the decision to re-lock the product would probably be given the same credibility as Mr. Reagan's claim not to   remember anything between 1981 and 1989.

Consequently, I would like a letter drafted and signed by you, then notarised, that states the following:

John Walker was not consulted in any manner regarding the decision to reimpose a hardware protection device on domestic copies of AutoCAD.

John Walker did not participate in any form in the deliberations prior to this decision, nor did he participate in the decision itself.

John Walker was not informed of the decision to reimpose a hardware protection device on domestic copies of AutoCAD before 1:56 P.M. PST on Monday, January 29, 1990, at the same time all other members of the AutoCAD software development group were informed of this decision.

In addition, John Walker has attended no senior management, board of directors, or equivalent meetings since June 1, 1989, and has not otherwise received the information presented at those meetings, nor participated in any decisions made there.

Capitulation

  January 30, 1990

The decision to reintroduce the hardware lock was not made without a great deal of time, thought, and heartache by management. I was on the front line of the last firestorm and am not anxious to hurl myself headlong into the snakepit again without good reason. O.K., now that I've said that, the reality is that this issue has, to say the least, attracted quite a lot of attention. The good news is that it's refreshing to know that the people at Autodesk still care extremely about the direction of the Company. The bad news is that already a great amount of time, thought and anguish are being given to one topic at a time when we are under the greatest pressure that we've ever had to get out an AutoCAD release. The hardware lock is not going to drive a wedge through the Company and I'm not going to risk the release of Version 11 through stubbornness. The lock will not go on Release 11 and the Thursday meeting to debate the issue is cancelled. Let's direct all of our energies to getting Release 11 out on time.

Alvar Green


Next Up Previous Contents Index
Next: ``Autodesk Arrogance'' Up: The Autodesk File Previous: Conclusion

Editor: John Walker