There has been much lamentation and gnashing of teeth in recent times about the evils of software patents. There is wild controversy about whether the whole idea is fatally flawed—for example, Dave Winer has argued that software patents are bad economics and (in the U.S. context) constitutionally unsound. Further, there is a widely-held belief that the US PTO has been too uncritical, and insufficiently attuned to prior art, in issuing such patents. Here’s a confession: I currently have two software patents in the US PTO pipeline, and did some work on them last week. Herewith some narrative of what the process is like from the inside, with commentary on the broader issues.

Patents, Startups, and VCs · Antarctica, which I founded and which employs me, has received several rounds of Venture Capital funding. When I sought the first round of investment, it was made clear to me that VCs take patents very seriously and that I would be required to apply for patents as a condition of getting the money. More on my feelings below, but since this seemed like a gate that I had to go through to give my Antarctica vision of a graphical user interface for everything a chance, I went through it.

How It’s Done · I and our VP Engineering Dave Ashworth filed two patents at the US PTO in the fall of 2000. A patent has three parts: a narrative, some illustrations, and claims. The narrative gives background and a (somewhat) human-readable explanation, the illustrations are illustrations, and the claims are where the meat of the patent is, explaining exactly what aspects of your invention you’re claiming. The claims are written in a version of English that is so fractured and stylized as to be essentially incomprehensible to most people.

Here’s how it works: you select your pattent attorney (we’re using Oppenheimer, who seem pretty good), and you do a bunch of searches to research prior art; mostly you hire specialists to do this part, because it’s pretty arcane. At one point we built a drop-dead beautiful Visual Net interface to a patent database which would have made it much easier for everyone to do this kind of searching, but oddly enough the specialists seem to like it the way it is.

You pay the patent lawyers a substantial chunk of money (somebody joked that it was $1,000 per claim), count on several tens of thousands for a couple of patents. They take your explanations and drawings and compose the claims. Then, you file the whole thing at the US PTO. Then, you wait.

As I said, we filed in late 2000 and they finally got around to looking at one of the patents this summer (one remains unlooked-at). When we inquired about the slow progress, they noted that filings in this particular area (information-technology patents) have been heavy and that the PTO is backlogged.

The lawyer said that quite likely the PTO would initially flatly reject the patent on the grounds of prior art; that this was no cause for concern, it was how the system worked, that it didn’t really change the likelihood of eventually getting your patent approved.

And that’s what happened to our first patent; the examiner sent back a note citing two other patents and explaining how each and every one of our fifty-odd claims was covered by one or both in combination. The lawyers said “See, just as we said.”

I’m kind of cynical and paranoid and it crossed my mind that it was awfully convenient for the lawyers to have this narrative in place if they wanted us to go on spending money on what might be a doomed patent. Barry Yates our CEO tracked down another senior intellectual-property lawyer through his personal network, who confirmed everything that the Oppenheimer folks had said, that this was typical and not that meaningful.

Paper Chase · So, last week, I cleared out a big chunk of desk space and sat down with our filing, the two other patents, and the examiner’s notes.

A few minutes later, I was on the phone to our lawyer. The examiner’s comments on our very first claim had cited language in one of the other patents (by column and line number, very precise) that covered our claim. Only when I went to that place in the other claim, they were talking about something completely different, I didn’t get the connection at all. So I wanted to know what was going on.

The lawyer pulled up the patents and pointed out that the sentence in the other patent used a word that was a close synonym for one of the key words in our claim. He explained that the examiners are insanely overworked and under huge pressure to get through the maximum number of claims every day, and (at least in this first-cut situation) may take an approach as simple as digging up other patents in the space and running through them with the PDF search function and a thesaurus. So if it seemed to me like the examiner was off-base, I was quite likely right.

This is not to say that the examiner is a dummy. First of all, she or he had dug up a couple of patents which were in a related space, and had made some non-obvious connections that I might not have thought of. She was also obviously quite technically competent and had no trouble with the terminology or math.

Having said that, on closer examination it was obvious that the two other patents were actually describing radically different inventions, and the overlap was mostly at the level of language rather than technical content. So we’re going to go on with the process, because whether or not you think that patenting software inventions is a good idea, I’m pretty confident that in our cases these are actually new inventions; this is a field I know pretty well. We’ll send a few more thousand dollars to the lawyers and try and get this thing done.

Is the Process Broken? · Obviously, it’s unacceptably slow; but on the other hand we can hardly give these people grief for not doing enough work on prior art and at the same time get mad at them for not going fast enough.

As to the quality of the review, I thought it was actually pretty good, if hurried. While I’m convinced that the invention really is distinctly new, the examiner’s remarks highlighted that our description of it failed to explain that in an understandable way; we’re just too close to the problem. As I’ve said before, when the person you’re explaining something to doesn’t get it, that’s usually your problem, not theirs.

I was initially a bit surprised that the prior-art search didn’t go outside the domain of patents. There has been lots of other work done in our space in both academic research and product development, and some of it was at least as close to our work as the patents the examiner cited. But I thought it over a bit and on consideration, the two patents the examiner highlighted were really good representatives of the work that’s generally been done in our field.

The effect, at the end of the day, was that I had to put a considerable chunk of work into thinking about why the invention really was unique and writing the results up carefully and thoroughly. Which strikes me as highly appropriate.

Given that many egregiously-silly patents have been issued in the software space, I was pleasantly surprised at the quality of the review that mine got. Either our examiner is sharper than average, or the PTO has significantly pulled up its socks.

Are Software Patents a Broken Idea? · I really don’t know. One of my brothers, an Industrial Designer, has his name on a patent for a device for mixing gases that’s used in chromatographs. When he showed me the filing, with the drawings and schematics and so on, I was impressed; these guys had cooked up a new arrangement of valves and geometries that did a practical task in an elegant and new way. It felt much more rigorous than the way we go about inventing new technology in the software space; but maybe that’s just because I’m way too close to the software world and can see all the warts on its underbelly.

I’m inclined to think there’s a spectrum of reasonability in software patents. “One-click ordering” seems like a grievous error, simply because if you said those three words to any web-savvy ecommerce-savvy programmer, they’d say “OK” and build it for you and it would work; which doesn’t seem to meet a high enough bar to qualify as an invention. But consider the basic PGP setup by Phil Zimmerman, it’s just immensely clever and elegant. I have the feeling that that really does qualify as an invention in totally the same sense as my brother’s gas-mixing apparatus. Obviously I think the things I filed are closer to PGP than one-click ordering.

As for Dave’s argument that software is protected from patenting on free-speech grounds, I’m not remotely competent to judge it but it would be fascinating to see that one get hashed out in court.

Are Software Patents Economically Sound? · No.

First, the nature of software is that the cost is all in development, not replication, but the patent model of intellectual-property licensing really works best when you can collect a royalty per-copy.

Second, the evidence seems pretty clear that at the macro scale, the economy benefits most from software when friction is minimized so that new things get out in the market quickly and morph and grow and are adapted and evolved and different groups compete on better execution. Patents are all about friction, because there are lawyers in the picture.

Third, patents are supposed to encourage innovation by rewarding inventors, but when it comes to software, that encouragement seems to be superfluous. People who have significant innovations to offer typically would have to be locked locked in a room with no computer access to keep them from cooking up cool new stuff and offering it to the world.

Finally, it’s obvious that Open Source is a normal healthy part of the software ecosystem, and things generally go better in areas where there are Open-Source alternatives. Patents and Open Source are violently incompatible.

Are Software Patents Good Business? · This is a hard question. In the sixteen years since I got far enough into management to see the business picture, I’ve never taken in a single penny of revenue on the basis of having a patent. I have never lost a single penny of revenue on the basis of not having a patent. I have however spent probably a couple of hundred thousand dollars over the years on intellectual-property lawyers’ bills. I wonder (I really wonder, this is not a rhetorical question) whether any software company in history has derived sufficient revenue from software-patent royalties to cover their legal bills? Put another way, if you put all the software-patent legal bills in one bucket and all the software-patent revenues in another, I wonder which would be bigger?

But, the experts explain, this isn’t about revenue, it’s about competition; if there’s a competitor you’re worried about, slap them with patent litigation and screw up their go-to-market execution. Or, if you’re not comfortable with doing that, what do you do when someone does it to you? One of the best strategies, when hit upside the head with patent-infringement litigation, is to have a portfolio of your own patents you can hit back with, so that you don’t get derailed, you just pay a bunch of legal bills and settle out of court. It seems to me that this generally sucks, but is nonetheless the best argument I’ve heard for why a startup company might want to invest in securing some patents.

And venture capitalists perceive that patents are tangible, objective evidence of real intellectual innovation; something that isn’t just a wild-eyed entrepreneur’s pipe dream. I’m quite sure they’re wrong; significant innovations are best distinguished by the fact that people are willing to pay money for them.

Conflict of Interest · The people in the system with the worst conflict of interest are the patent attorneys and various flavors of intellectual-property expert, who are always willing to argue passionately and at great length about the importance of securing patents, and how this is an essential part of any business strategy. They seem to really believe what they’re saying, but when I listen to them my mind’s eye just can’t help seeing the fat envelopes stuffed with carefully-itemized five-figure legal bills crossing my desktop.

A cynic would say it’s a self-perpetuating system whose costs greatly outweigh its benefits, and that we’re hopelessly stuck in it. But then, I’m a cynic.


author · Dad · software · colophon · rights
picture of the day
September 15, 2003
· Business (106 fragments)
· · Intellectual Property (48 more)
· Antarctica (17 more)

By .

I am an employee
of Amazon.com, but
the opinions expressed here
are my own, and no other party
necessarily agrees with them.

A full disclosure of my
professional interests is
on the author page.