Removing the Hardware Lock

With the introduction of AutoCAD release 2.1, all versions sold outside the United States and Canada were protected by the “hardware lock” or “WIDGET” (Walker's Inline Device Guaranteeing Elimination of Theft). This is a transparent RS-232 device which AutoCAD probes and requires to be present in order to run. When the introduction of this device went reasonably smoothly, Autodesk U.S. introduced it in the domestic market in release 2.5 in June of 1985. Never in our wildest imagination could we have anticipated the reaction. Suddenly we were exposed to a blast of vilification, moralism, and hypocrisy that (in my case, at least) forever changed the way I'll approach selling productivity tools to customers. All of the industry analysts and press people who had questioned us sharply about the threat of piracy in our market abandoned us to take the heat of trying to do something about it alone; not one word of support for us was written. Competitors jumped in to promote their products as better because they did not prevent theft, and products appeared on the market which claimed to defeat our lock, and were marketed “only to benefit the customer”. The most notable of these products was itself copy protected.

On September 19, 1986 I recommended that we remove the hardware lock from the domestic product. After extensive discussion and preparation, we announced that the lock was being removed on November 25, 1986. The following document was distributed to all Autodesk employees on November 25; other than talking about the issue as a firm decision instead of as a recommendation, it is identical to the original memo I wrote to management recommending that we pull the lock.

We have continued to use the lock on international versions, and have encountered none of the problems we had in the U.S.

I think the time has come to admit that we made a misjudgement with respect to the hardware lock and plan an orderly retreat from the mess we have gotten ourselves into.

This paper will try avoid issues of morality and justice and focus on the business issues involved. I think we have all hashed over and debated the morality of this to exhaustion and in fact that's one of the reasons I make this recommendation. I will only reaffirm that I continue to believe that our software licensing policy of one user, one license is the only sound foundation on which a viable software industry can be founded; that the fundamental problem with the hardware lock is that it prevents theft of our intellectual property; and that any moral opprobrium we direct at those who steal the product of our labours is fully shared by the manufacturers of computers who profit by selling machines which provide no form of protection for the property which makes them useful.

I think that we underestimated the hypocrisy, moralism, and disdain for intellectual property that exists in the United States. It is not enough that we provide ever-increasing functionality at incremental update prices tiny by comparison with any other industry; any attempt to fund ongoing development through incremental sales is seen as a “large, rich company” oppressing its small, struggling customers, the overwhelming percentage of whom signed or implicitly assented to a license agreement which our hardware lock only acts to enforce. We must not only tolerate looting, we must not attempt to prevent it. Ayn Rand called it “the sanction of the victim”.

But we must recognise that we are only a software company, not a major force for morality in the world. We must make the decisions which will make our company and our products prosper and try to act in the right within the constraints of the real world. And it is on that basis that I base my argument here. I think that there is nothing we can do in the short term or medium turn to reverse the moral climate which opposes us presently. Only a long term shift in perceptions, aided by a concerted, united effort by all software providers and supported by hardware manufacturers (a signal example of which would be IBM, Compaq, and Apple pledging that all new PCs made after 1987 would contain a serial number chip) can help. A climate where falling hardware prices is presumed to cause lower software prices is one in which much education remains to be done.

The issue of intellectual property protection only seems to respond to this type of coordinated fix. It was not a revulsion with cassettes, a fee per tape or tape deck, or the FBI raiding pirate pressing plants that the music industry finally settled on as the solution to its problems: it took a new, uncopyable medium, controlled by a strict licensing mechanism and a high capital start-up cost, and a pledge by the hardware manufacturers to forgo revenue from making a medium permitting direct copies (the 44 Khz DAT agreement).

In implementing the hardware lock with release 2.5, I think we not only misread the moral climate in the U.S., but we also made two public relations errors. First, we failed to get out front and promote the hardware lock as license enforcement without copy protection. In retrospect we should have hit the ground selling; explaining how the lock was central to the vitality of a local dealer and support channel, a part of our ongoing commitment to R&D and low cost updating of installed customers, and the alternative to the large-corporations-only site license mentality so many other software companies are adopting which ends up discriminating against the little guy who built this business.

Second, we failed to sell 2.5 as the massive update it was. It was always our intent to “spring” the hardware lock on a release so compelling that all users would be forced to upgrade. 2.5 is such a release, but we have not sold our customers on that fact. The demand for the DXF downgrade program is the most evident symptom of this fact. And of course our underpromotion of 2.5 was compounded by the focus on the hardware lock in most of the coverage of it.

I believe that it is too late to remedy either of these errors now. One, the discussion has become so polarised over the hardware lock, 2.5 is considered “the copy protected AutoCAD”, and we would have to overcome a very high barrier of resentment just to be heard. Two, our key users already have 2.5, the reviews are in, and they are well into the phase of picking it apart, preparing wish and complaint lists, and looking to what Autodesk will do next. A major push to promote 2.5 as a new release would look odd and defensive at this date (I distinguish general promotion of AutoCAD as it stands, the central theme of all of our promotion which should continue and be expanded, from specific “new and improved” promotion aimed at the installed base and at industry decision makers).

I think the central issues here are bad faith and good will. The small business users on whom our success has been based are guilty of what can only be called stunning hypocrisy and bad faith when they install additional computers at five to fifteen thousand dollars each but claim that theft of additional AutoCADs is the margin that keeps them out of bankruptcy. However, this market is the heart and the soul of our business and we should decide, and soon, if we want to debate with it or sell to it. Autodesk has over the last four years, accumulated a large reservoir of good will, respect, and trust among the small user community. Our dealer channel is successful largely because of this market. To the small user far more than any other, AutoCAD is CAD, and Autodesk is seen as the company on his side, as opposed to an IBM, Lockheed, or McDonnell-Douglas. We seem to be spending this good will at an extraordinary rate, and purchasing very little with it.

And that's the final argument: we've been accused to forsaking our ideals and focusing on “the bottom line”. All right, let's do some of that. We have just completed the best quarter in the company's history, but we have always had a large bulge in sales after a new release. I would be very hard pressed to argue that hardware-lock-induced additional sales have contributed much to this sales performance, versus the normal post-new-release bulge we've experienced. There is a convincing counter-argument to this: that due to the hardware lock, our installed base has purchased only their first low-cost 2.5 upgrade. When, over the next 6 months they actually experience how useful 2.5 is, they will find the money to legitimise their additional copies and we will reap those sales. Were I an AutoCAD user today, I might well decide to wait and see if Autodesk crumbled under the pressure and removed the lock rather than ponying up $2750 each for my bootleg copies.

Having said all of that, my recommendation is based on these simple facts: we have not experienced a large increase in sales based on shipping the hardware lock; we are expending at a rapid rate and with little obvious return the good will of those most satisfied with our products and most influential in recommending additional purchases; we have failed to find strong support for the hardware lock among the very dealer community which is most benefited by it; our precious management, technical, and product resources have been diverted into a largely defensive effort; we are imperiling our perception in the market sector we most control at the very time that our lack of obvious technological leadership and growing competition from larger vendors puts our future most in doubt in those corporate and government accounts least likely to be worried about the hardware lock.

We should remove the lock now, on an incremental 2.5 update, because if we wait until the shipment of 2.6, all the publicity attendant upon that release will be buried under the news of our removing the lock. We lost almost all of the publicity on 2.5 enhancements in the debate over the lock, and we simply cannot afford to have the news of 2.6 buried in news of our reversal. We should concurrently go on the offensive with a promotion campaign explaining what we have done and why. This will act to palliate the inevitable blast of “Autodesk repents major marketing blunder” publicity which will attend our announcement.

I expect that the two weeks after our removing the lock will be very difficult weeks. I expect those who said that they would re-embrace us as the market leader if we removed the lock will remain silent, while those moralistic mountebanks who have been reaping profits larger than ours by far as a percentage of sales by selling products purporting to “break the lock” will crow over their “victory”. Further, I expect some of the very dealers who have been silent or petulant about the lock will now view its removal as an assault by Autodesk on the viability of their businesses. And we will be assailed by publicity and cheap shots about our “blunder”, “indecision” and the “shakeups in Autodesk”. One of the principles I've always followed in business is that there's nothing wrong with being wrong—if you never try something that entails risk you're doomed to stagnation and eventual failure. Catastrophe is engendered by staying wrong in the face of clear evidence that you're on the wrong course. I think that we're far better off putting this episode behind us now. I believe that we are doing the right thing in getting this over with and getting back to what we do best: developing, selling, and supporting products which revolutionise the way designers do their work.

This is a prototype of the original (version 1) hardware lock, assembled from the first run of printed circuit boards. The hardware lock worked by applying a Boolean function to a challenge sent by the computer, and verifying the output. The jumpers on the board allowed setting the inputs to the function. The actual hardware locks shipped with AutoCAD were encapsulated in plastic.

The hardware lock was subsequently redesigned to be smaller, easier to manufacture, and less expensive. The ability to specify the inputs to the function with jumpers, which was never used, was deleted. Again, this is a prototype; production hardware locks were potted in plastic.