How to Publish a Patent

Patents are neither mysterious nor lofty. A patent is simply a public technical method for solving a problem. In my view, the truly lofty patents—the ones that deserve the name—are those that advance human civilization (changing, or about to change, how we live and the habits we keep): Apple’s multi-touch, Tesla’s battery technology, even Twitter’s pull-to-refresh.

Why Do Companies Publish Patents?

By purpose, patents generally fall into two categories:

• Self-protection for a “killer” technology used in a product, such as multi-touch.
• Patents filed specifically to round out one’s patent portfolio.

To get a patent of the first kind, unless you build a breakthrough product and play a leading role in it, forget it. Let us set that category aside for now. The second category splits again into two cases:

• Offensive patents. Examples include Qualcomm’s patent portfolio around CDMA and Microsoft’s portfolio around handheld devices. If you want to build a product in that domain, you cannot avoid my patents; so either pay up and swear allegiance, or follow my lead.
• Defensive patents. Probably 80% of the world’s patents fall into this category (mine certainly do). How does the defense work? There is a story in the IT world (impossible to verify): IBN had clearly infringed some of SUM’s patents, and SUM prepared to sue IBN. IBN’s patent lawyer patted SUM’s lawyer on the shoulder and said, “Buddy, when you were still in diapers, we were already the big boss on this turf, with countless patents in hand. We’re all in the same business; we see each other all the time. If you push us too far, we’ll go back and dig carefully through our patents—don’t tell me you haven’t infringed on our territory. Let’s eat together and drink the soup together; wouldn’t that be nice?” SUM’s lawyer trembled, and the matter was settled. Why did Google pay a fortune for Motorola but keep only its patent portfolio and throw away the rest? To fight a thug, you first need some thug credentials yourself.

How Should I Publish a Patent?

I’ve rambled too much. Back to the point—how to publish a patent. I work at a foreign company, so the patents I have filed are all U.S. patents. What follows may differ slightly from domestic Chinese patents (but not by much).

A patent must satisfy two all-important, universally applicable properties:

• New
• Non-obvious

Novelty is easy enough to understand: others don’t have it, you do—that is novelty. Non-obvious means the thing is the product of thought, not something that is obvious at a glance. If you wanted to patent a method of eating, “pick up utensils, pick up food, open mouth, eat” is obvious, and you probably cannot patent it; but “to eat hand-held mutton, use a small serrated knife to cut at an angle, like this, this, and this,” plus a drawing of the specific cutting technique, and you have a good shot at a patent.

More concrete? All right. These two requirements are too simple and too subjective (you think it’s obvious, but my IQ is low and I don’t agree—what then?), so detailed rules are needed to guide the process. Generally there are points like these (possibly wrong, possibly incomplete):

• How important is the problem the patent solves to the product? The more important, the more it needs to be filed.
• How much can the patent protect us? The stronger the protection, the more it needs to be filed.
• How much can the patent be used to strike at competitors? Or, how likely is a competitor to infringe our patent? The greater the likelihood, the more it needs to be filed.
• Is the patent within the company’s patent portfolio? If you work at an oil company and want to publish a patent on app pull-to-refresh, forget it!
• …

Such rules differ by industry and even by company, but they are broadly similar (different paths to the same end). Just as athletes need to know the rules of the game before they can achieve the best possible result, if you want to publish patents, you first need to understand your company’s patent review rules. Generally, every company that has established itself in an industry and plans to stick around has a patent (technology) committee. You write the patent; they review it. They score it using questions like those above—very subjectively, just like Olympic gymnastics judges.

As programmers, once we understand these rules, we need worry about only one thing: is the method I used in such-and-such product novel enough?

You may ask, what about the non-obvious requirement? You really don’t need to worry about that. We work in software; it is extremely, extremely brain-intensive, so surely it’s not obvious! So the key question is whether someone has already used a method basically similar to yours. But proving that requires searching through an ocean of patents—far too hard. Anyone who has read patents knows that reading them constipates you more than reading academic papers, because they are not written for normal human beings at all!

Don’t panic. Have some confidence. If your method did not come from Google (congratulations), then Google it and see whether you can find something similar. If not, you’re broadly fine. Leave the rest to your company’s patent committee. They will set up a Hongmen banquet (a trap disguised as a feast) to give you a hard time, and you just need to talk your way through it.

One of my patents, a hot patch generator, was given exactly this treatment. A big shot on the technology committee found a Java VM JIT-like “Monkey Patch” method and said, “Look, programmer, Java has already done this.” Fortunately, the key point of my patent was how to generate a hot patch by comparing the two binary files before and after modification. The emphasis was on generating the patch, not applying it, so I eventually convinced the committee and made it into the formal application process.

From the moment you have an idea to final approval, a patent takes three or four years if all goes smoothly. Generally the process goes like this:

product -> abstract the patentable method from the product -> internal review (company patent committee) -> write it with a patent attorney -> external review (patent office) -> national publication -> final approval

If you pass internal review, you’ve basically finished the game. The rest depends on your company’s capability and influence. The more patents a company publishes, the greater its relative influence (just like the Chinese gymnastics team’s influence with the International Gymnastics Federation). The company will assign a patent attorney to work with you on the patent. You cannot write the actual patent; that is a word game for lawyers. You tell them your documents and ideas in exhaustive detail; they do the writing. When the draft is done, they will ask you to review it, but frankly, aside from a few figures where you can barely make out your own hand, the text is unrecognizable.

Once a patent enters external review, you can only wait passively. One patent I filed in 2008 was only finally approved in 2013. In between there may be several rounds of back-and-forth (government efficiency, you know how it is), a pile of useful and useless citations added, the scope of application revised countless times, and finally publication (you will have no idea about these changes; they all happen between the patent office and the lawyers).

Yes, the complete process of filing a patent is enough that you could damn well fit in another undergraduate degree (in medicine, no less).

What Are the Benefits of Publishing Patents?

First, every company pays a generous reward for patents that pass internal review, roughly 5%–20% of the total patent application cost. The company I work for is very generous on this point. If you are short on cash, this is a good opportunity to increase your annual income, and your personal market value will grow, further increasing your annual income. After buying a house in 2008, I suddenly became a “millionaire in the red.” Selling my body or a kidney was not an option (the iPhone wasn’t that hot yet), so I had to sell my craft instead. So in those two years I cranked out four or five patents (and made a decent pile of money); now the harvest season has come around again, and one patent after another is being formally approved (and starting to burnish my résumé).

Second, this is a good opportunity the company gives you (if it does) to enrich your résumé. Not seizing it would be a waste of heaven’s gifts. Generally, the total cost of a U.S. patent up to publication is about $10,000–$20,000 (see How Much Does a Patent Cost?); domestic Chinese patents probably cost no less, mostly on lawyers, because you have to hire them to write it (think of the pittance we get for writing articles—so pathetic). So if you want to file on your own to burnish your résumé, forget it. You can’t afford to play, and you can’t afford the time.

So while you’re young, if your company has such opportunities and processes, hurry up and look for those little flashes of brilliance in the products you work on. Think about them more in your daily work; maybe you’ll bag one.

Of course, for programmers, shipping excellent products is the real priority, but a little icing on the cake like this is, as Han Xin said of troops, the more the better. It’s like publishing papers in core journals: once you’ve written one, you know it’s no big deal, but if an undergraduate publishes a few more, they always get a head start in the job market. Right?