Monday, October 08, 2012

The evil engineer's guide to patents

tl;dr: It's super-easy to file software patents and boost your resume, if you don't care too much about doing evil.

You see the following in the bio of a speaker at a software conference: "PhD from Stanford, built three successful startups, owns four patents." Envious? I can't help you with the first two, but the last one's easy. Here's how you too can boost your resume by dragging down human progress.

As for me, I'm the inventor of 13 patent applications. So if I'm about to say anything unpleasant about pumping up your CV with patents, at least it's not sour grapes. Pretty good, considering, from a man named Fox.In 1999, I invented COMET, the Web protocol for pushing data to a browser. COMET is widely used and is now being made part of Web standards. Yes, I invented COMET: Look here, a real patent filing from 2000. Any experienced software engineer will tell you that this software design has been used for decades. No, I didn’t invent COMET. But because I invented COMET -- I rock! I'm a rockstar ninja! Hire me for a big bundle of money! 

If you are moderately smart (and if you’re not, stop reading this), you too can be in on the scam. The average time needed per patent is roughly 15 work-hours.
You can then brag, truthfully, that you own the patents. Your employer has all rights to them, but as you'll soon see, that doesn't matter much.

You do not need a new idea, just a good enough idea for convincing your bosses to file it. (If you're going it alone, you don't even need that.) Brainstorm a little, think of a cool idea, then search the web and Google patents for half an hour. If you don’t find your idea mentioned exactly, you can probably patent it. All well-known ideas in software have dozens of patents on them, so it's not a problem if the idea already exists.

You don't have to write up the patent. All you need is to describe it to a patent attorney, orally or briefly in writing. The attorney will multiply the verbiage by about 40, writing in jargon which is carefully designed not only to broaden the patent, but also to confuse the underpaid patent examiners so that they approve your patent. (One purpose of patents is to share your idea with other engineers, but you will not understand your own patent after it's written.)

I won't go into detail on how the patent system really works. This is a guide to "inventing" for working grunts like you and me. Please read up on the quagmire that is the intellectual monopoly system. But the first thing to know is that software patents are not used to protect a company's innovations, nor to sell license to use the invention. Instead, they are used to threaten other companies with Mutually Assured Destruction. As in “Wanna mess with me? We have 50,000 patents in our files, and I bet that if we check hard enough, you're violating one. Back off, sucka!” Patents are also used to shake down companies for protection money--particularly those who appear to have enough money to extort but not so much that they can defend themselves.

You don’t care much if the patent office accepts your patent. What's important to you as employee inventor or seed-stage founder is that it gets filed. You can honestly list "patent applications” on your CV or investor datasheet. After that, it takes five to eight years for the patent to get finally approved. Where will your career/startup be by then? 

A pop quiz question: Is it better to patent a deep, insightful, revolutionary new algorithm, or a trivial new GUI widget? If you guessed “cheap GUI tricks,” you're well on your way to filing your first invention. The more obvious a patent is to a layperson, the less likely it is your pointy-haired bosses or corporate patent committee will write it off as weird boffin stuff. Also, it's easier to threaten to sue other companies for what can be seen in the GUI, rather than for deep-down algorithms that no one can see, nor understand if they could.

To get your bosses to approve your idea for filing, get your product manager to say that the idea is in the product roadmap. (A product manager will put anything in the roadmap -- for the year 2037.) Then, you can say that the patent will protect features officially specified for a future release of the product.

If you're asking, "But what about innovativeness, non-obviousness, business value?" then, good: You've read up on what counts in a patent. You need that knowledge to word your patent right, so that your boss or patent committee agrees to file it. Start thinking like a lawyer: Your goal is not the truth, but rather winning, within the limits of professional honesty and ethics.

Here's a way to become your colleagues' best buddy: List your cow-orkers as co-inventors. They do the same for you, and everyone gets more patents on their resume. If you manage a team, you can do even better: Get yourself listed as an inventor -- your underlings won't argue -- then make them do the scut work of pushing it through the filing process.

Here's how you add people as inventors: Have a brainstorm meeting to refine the idea. Any participant with half a brain (sorry, managers, you can only pull off this trick if you have at least half your brain left), will likely contribute something significant to the discussion. When they do, one not only may, one must list these contributors as co-inventors.

Don’t worry if the colleagues didn't help you slave over steaming Bunsen burners, while the Tesla coils hiss and sparkle in the background. Not required by law.

Still think a patent needs to be a new idea? There is an official, approved, signed-and-sealed patent for trimming whitespace when inserting into a database -- that means, converting " John Brown "  to "John Brown". It's from 2009, and engineers have been doing this rather dull little piece of database code for 50 years.

No, wait, stay tuned for more! There’s another patent pending from the same inventors, this one for trimming white-space when selecting from a database. Two patents on their tresume instead of one. Come on now, you’re smart enough to come up with ideas like that, aren't you now?

I once asked an experienced inventor how I could make money off a patent, if I actually were to control one myself. Clearly no one is going to pay me royalties for the privilege of using my yellow web-widget. This is his answer: First, talk up your yellow web-widgets in conferences and articles. Meet with companies and try to convince them that yellow web-widgets are the coolest thing since liquid helium. Don’t mention that you have a patent. Then, when they develop a yellow web-widget, sue 'em ‘til they bleed. (The term  "patent ambush" is used for doing this while hiding patents in standards.)

This guide is mostly for employees, but if you're doing patents on your own, here's how to start off cheap. File a provisional patent in the US (the only country that counts) for $110, with a brief description in ordinary language. It lasts for a year, and you can file up to a year after you release your “invention” in a software product (if you even intend to do that). So, you have two years to find funding for the real patent, or just to abandon the provisional patent once your company is either stable and successful or stable and dead.
Try not to implement your patent idea. All you need for filing is a vague description, and if you decide to patent a feature which is in your product, you'll just be under pressure to file by the deadline.

Not that it matters, but here is an aside on what happens after six or so years when the patent examiner (some huge percentage of which are Vietnamese and Ethiopian immigrants, for some reason), looks at your patent. He can't understand what it says, of course, so he searches his patent database for some keywords. Because your patent attorney obfuscated it, there are no meaningful words that relate to what your patent was originally meant to do, so he end up searching for contentless words like "display information on a screen." 

Naturally, he finds a patent with those words but which has nothing to do with your original idea, then sends your application to what's called "final rejection." If he gives you a chance to argue first, you are not allowed to say that the prior patent is different from your original idea. (It is, though there are other patents which are functionally identical to yours.) Instead, you need to pick away at the contentless keywords like "display information on a screen," and explain that there is some miniscule difference in the act of displaying information on a screen. This makes no sense.But anyway, most patents applications are in in the end approved, though usually after many rounds of "final rejection" and further wrangling, each of which enriches the patent attorneys and the USPTO. (At this stage, sometimes applications are sliced into multiple applications. Yay, another patent for you! I, for example, own two patents with the same name.)

In the very rare cases that applications fail to turn into patents, it is usually for silly reasons like the bankruptcy of the company you worked at. No sense worrying about that when writing the patent. In any case, though approved patents get listed as such on Google Patents and the USPTO site, failed or abandoned applications do not, so you will never have any reason to remove them from your resume.

Now, we'll take a break while you bang out a patent.

. . . Go ahead, it's fun to get official government approval that you are a true innovator . . .
After filing, list the applications on your CV or personal website; then, write a blog post denouncing software patents.

Next, if you’re looking to salve your wounded conscience, try the same rationalization used by nuclear weapons engineers: Without the threat of Mutually Assured Destruction, where would we be now?

Or maybe the world has enough sense to get rid of this negative-sum game? It might take a while. While you’re waiting, read Against Intellectual Monopoly, by Michele Boldrin and David K. Levine, Cambridge University Press, 2008. You can download it here. Or just read this article at The Atlantic.


1. This blog is most definitely my personal opinion and does not reflect the opinion of any current or former employer.
2. I Am Not A Lawyer. The article may include mistakes about the patent process. I also made some generalizations for the sake of brevity. Figure it out for yourself.
3. The above is written from the frog's-eye perspective of an software engineer in the trenches. There is also the bird's-eye perspective of a large-company executive, which is quite different, though just as grotty.


Ray Lopez said...

So, what is your position on novelty vs non-obviousness? Is a patent that claims "single click" for the first time patentable or not? Before you answer, keep in mind everything is obvious "after the fact". Einstein's Special Theory of Relativity was obvious in view of the Lorenz Transforms but ONLY after the insight that nothing travels faster than the speed of light. Granted we are mixing metaphors as you cannot patent a theory, but you can see the problem with your simplistic engineer's approach. As for innovation, it proceeds in stages. Everything invented now will be obsolete in 100 years, but that's not an argument for not patenting anything. Patents are stepping stones.

Joshua Fox said...

@Ray, from the evil perspective of this article, it doesn't matter whether the patent is obvious, novel, or anything else. All that matters is whether you can file it.

What I really think is that software should not be patentable.