Fourmilab Home Page

Talkin' 'Bout My Innovation


Having been burned and forced to pay off a patent troll who claimed to have invented and obtained a U.S. patent on the use of exclusive or to nondestructively draw a cursor on a screen (a common practice in computer graphics for years prior to the purported “invention”) during Autodesk's Initial Public Offering in 1985, it was dismaying to see abuse of software patents continue to rise in subsequent years. It seemed that patent examiners simply didn't know very much about software or were insufficiently diligent in researching prior art, resulting in a situation where it was relatively easy to obtain a patent on a technique which had been in widespread use in the industry for many years, and then use the patent to extort license fees from those who had been using it long before the patent application was filed. This seemed to risk an outcome where the free-wheeling innovation which characterised the software business might seize up into an endless tangle of litigation where every time you wrote some code you needed a quarrel of lawyers looking over your shoulder lest you use an algorithm which somebody else had patented.

The only defence against this seemed to be to amass your own portfolio of patents so that if somebody came after you, you could find one of your own patents they infringed, and then offer to cross-license your patents and theirs. IBM and AT&T were masters of this strategy, with huge portfolios of patents generated by their research laboratories, which they used to obtain access to patented technologies they coveted through cross-licensing. By the end of the 1980s, lawyers were advising Autodesk to patent everything we happened to think of to build a defensive patent portfolio. (Of course, doing this generated billings for the lawyers, so they weren't disinterested parties in making this recommendation, but it did seem to be the strategy many other technology companies were adopting at the time.)

I played my part by dusting off an idea I'd been kicking around since 1967, and had recently taken the first steps to integrate into AutoCAD: computation with dimensioned quantities. In August 1989 I wrote up a disclosure document as the first step in the patent application and dropped it into the hopper. From that point, the process was entirely hands-off—the patent lawyers handled everything without getting back to me. Finally, in October 1993, U.S. patent 5,253,193 was granted. Here are the original disclosure and the patent as issued. It's interesting to see how the statement of an idea is transformed into the specialised jargon of patent prose.

In September 1990, having implemented a preliminary version of this technique for AutoCAD, I wrote a second disclosure about the specification of units, including a system of units based upon the fundamental constants of nature (similar to the Planck units). This would have been a more complicated patent, as it included a language for defining units, a database of unit specifications, a means of compiling a set of unit definitions into the internal form, and the representation of units in terms of the fundamental constants. Autodesk opted not to pursue a patent based on this disclosure.