Lawyers, Guns, and Money: An Evening with EDAC on Patents

Next Thursday, October 29, 2015, the Electronic Design Automation Consortium (EDAC) is organizing an evening on patents and patent litigation. This is partially due to the US Patent and Trademark Office opening a Silicon Valley office in San Jose. This is actually a West Coast regional office, serving California, Nevada, Oregon, Washington, Arizona, Alaska, and Hawaii. They are there mainly to offer advice. In particular, their website takes pains to point out that they do not handle official correspondence (such as patent submissions), which are only done by mail or online.

The event is at the EDAC/SEMI offices at 3081 Zanker Road (the corner of Zanker and Montague Expressway). The evening has been organized by the EDAC Emerging Companies Committee, which in practice means Steve Pollock. I'm not sure how many emerging EDA companies there are these days. The event is a panel session but the only EDA company represented on the panel is Cadence, and I think we've already emerged.

The mayor of San Jose, Sam Liccardo, will provide a special welcome. The panel session is moderated by Salumeh Loesch of Klarquist Sparkman, with panelists John Cabeca of the San Jose USPTO, Karna Nisewaner of Cadence, Robert Sacks of Fenwick and West, and John Vandenberg of Klarquist Sparkman.

I have a somewhat jaundiced view about patents. When I started at VLSI Technology in the early 1980s, we didn't really have a patent program, we were too busy creating what came to be known as ASIC. Once we got large enough, companies like IBM, AT&T, and Texas Instruments came and dropped off a list of a few hundred semiconductor manufacturing patents and told us we were infringing some of them. We almost certainly were, but neither we nor the other companies had a clue which patents we might be infringing. You can't manufacture a semiconductor wafer without infringing patents, some for seemingly obvious things like filtering the photoresist. After some arm-wrestling, we ended up needing to pay some percentage of our revenue to all these companies as a patent royalty.

Note that we didn't actually "use" any of these patents, in the sense that we read them and followed the instructions in the disclosure. The value proposition of a patent that trades a temporary monopoly for disclosing all the secrets doesn't really work in practice, at least in software, semiconductors, and any other market I'm familiar with. In fact our legal counsel told us never to read patents since that way we could never be credibly accused of "knowingly" infringing them (which carries triple damages), just of unknowingly infringing them (single damages).

The CEO of VLSI hired a patent counsel and kicked off a major activity to build up our own patent portfolio. Since at the time we had some of the most leading-edge design tools, many of these patents were in EDA. Gradually our list of patents grew. In future years when IBM and co came calling, we had our own list of patents that they were probably infringing and so the royalty we had to pay was reduced or went away and we cross-licensed everything. That strategy doesn't work, however, with what are officially called non-practicing entities or unofficially patent trolls. Since they are typically companies just set up to own one or more patents, they are not infringing any patents and have no interest in cross licensing. Back then that wasn't the problem that it seems to have become today.

A lot of activity went into creating patents, both where I was working and in the older semiconductor companies. But these patents were mostly defensive, created to give us negotiating leverage. If patents didn't exist, we would have created exactly the same software and semiconductors and we would have wasted less effort on creating patents that nobody was ever going to read. Patents seem to work satisfactorily in a few areas like drug discovery where a single patent really can contain the "secret", the actual molecule. But for industries where there are thousands of small incremental improvements all the time, like semiconductors or EDA or smartphones, it becomes impossible to build a product without infringing those thousands of patents, and nobody has the time to read them, let along negotiate licensing agreements. Instead of oiling the wheels of innovation, patents seem to be gumming them up.

In the early 1990s, VLSI Technology got heavily involved in second-generation cell phones, GSM (which originally stood for Groupe Spécial Mobile and pronounced in French, and then got renamed as Global System for Mobile Communication in English). The GSM organization had the concept of essential and non-essential patents. An essential patent was one that you couldn't avoid infringing since it was written into the standard. Back then there was a lot less awareness of patents getting incorporated into standards, but now it is seen as very important, especially since the Rambus cases over DDR in the early 2000s. GSM phones were one of the first markets with lots of patents and high dollar volume and so a market where it was worth patent holders going after all the manufacturers. For example, you couldn't build a GSM phone without paying $1 to license the vocoder since Philips had, probably by accident, designed and patented the specific DSP algorithm that was picked. There were probably another dozen companies that owned key fundamental technology used in GSM.

At VLSI, we built chips and so typically did not infringe the patents directly since they were all patents for a cell phone (or more likely some of that special patent language like "bidirectional radio telecommunication device"). However, our customers using the chips would be infringing since the chips contained, for example, a compliant vocoder. Incidentally, I think this is also why nobody sues Google about Android patents. The patents are for a smartphone and Google produces an operating system. So it is companies like Samsung, Huawei and Xiaomi who infringe, by running Android on one of those bidirectional radio telecommunication devices. But another thing I learned from our patent counsel was that you always go as far down the value chain as you can. It is much easier to get $1 out of someone selling a $500 phone than someone selling a $25 integrated circuit. So it would make more sense to license the phone manufacturers than Google in any case.

The EDAC event starts at 6pm and the panel proper starts at 7pm on Thursday, October 29, 2015. See you there.