14 August 2011

Patents: Just Do the Maths

As I've noted, there is an sudden efflorescence of writing about the ills of the patent system. Obviously, on one level, that's great, but it's also becoming a little, er, boring. It means there are no contrary ideas to engage with, and that's dangerous for the health of the discussion, I think.

So I was really delighted to come across this post:

In the past few months, this rhetoric has grown to a furious roar, as the patent system seems to be affecting more and more of the technology industry in a negative way: small mobile app developers have been targeted with spurious lawsuits from companies that make nothing, major players like Apple, HTC, and Samsung are locked in patent-related litigation, and a pair of multibillion-dollar patent auctions has sparked an unprecedented war of words between Microsoft and Google. The most passionate critics loudly argue that whatever benefits our current patent system might offer have now been exceeded by its costs; that resources that should otherwise go to the development of new ideas are instead being misspent on the overzealous protection of the old.

This line of thinking has been so forcefully and insistently repeated that it has become almost axiomatic, an intellectual and rhetorical cheat that is rarely (if ever) questioned. But it’s also wrong — painfully wrong, in ways that sabotage any real attempt at reform. Being loud and angry is a great way to get attention, but it’s a terrible way to actually get anything done — especially since most of the emphatic chest-pounding sounds like a slightly dumber version of an argument we’ve been having in this country since Thomas Jefferson was appointed the first head of the Patent Office.

Splendid stuff - totally wrong, but splendid.

The article really makes two big claims. I'll address the second of them first, since it's more specific, and then look at the more general argument used.

If “the patent system is broken” is a lazy rhetorical cheat, then “software patents shouldn’t be allowed” is the most completely vacuous intellectual cop-out possible. The problem isn’t software patents — the problem is that software patents don’t actually exist.

What we keep calling “software patents” are just regular old patents; there is no special section of Title 35 that specifically delineates between hardware and software, or software and machinery, or software and anything else you might dream up. I don’t know when it became fashionable to pretend software patents were some funky and terrible new phenomenon, but it hasn’t always been this way: Y Combinator co-founder Paul Graham’s 2006 essay “Are Software Patents Evil?” remains one of the best holistic analyses of the software patent issue I’ve ever read, and it opens with “if you’re against software patents, you’re against patents in general.”

Well, yes, being against patents in general is certainly my position, but I don't think the author was looking for that kind of response. Instead, I'll explore his specific argument why software patents are just like any other kind:

But look a little closer and it’s easy to see that the boundaries between “just math” and “patentable invention” are pretty fuzzy. Every invention is “just math” when it comes right down to it — traditional mechanical inventions are really just the physical embodiments of specific algorithms. Consider the TurboTap long-necked draft beer nozzle, which was developed by a University of Wisconsin student named Matthew Younkle and granted US patent #7,040,359 — it pours beer faster and with less foam because of its long shape and internal structure. (I’ve conducted extensive… testing.) Isn’t that just a clever application of fluid dynamics? Where do you draw the line between the math that enables the invention and the invention itself? These aren’t easy questions, and we’re just talking about a beer tap. Things get even fuzzier when it comes to software, which doesn’t have a physical component to comfort our sense of justice. It really is just a bunch of math.

Now, speaking as a mathematician, I certainly concur with the view that everything is "just maths" in a certain deep sense: that is, we believe that we can, *in theory*, use maths to describe anything that exists. But in practice, some bits are trickier than others.

For example, take that TurboTap. As the author rightly notes, this is a "clever application of fluid dynamics" - except that it isn't. Fluid dynamics is one of those inconvenient bits of maths that we can't generally solve: the equations are just too complicated. Maybe one day some clever mathematician will come along with a clever method that will allow us to solve this stuff exactly, but I'm not holding my breath.

So what does this mean for the TurboTap? It means we can't design it using maths, because the instabilities of turbulent flow - which is pretty much all real-life flow - can't be calculated exactly. So the only way to design a TurboTap is to make lots of them, and find out which works best. In other words, you really have to *invent* the thing, because it's not discoverable from maths alone.

The same is not true of software. Although there are deep issues of checking whether programs work, the logic of a computer program is cast-iron: there are no fuzzy bits due to turbulence. If you simply apply the laws of logic and maths, you know exactly what will come out at the other end. So you are not inventing, you are discovering: those structures were always implicit in - and limited by - the rules of logic and maths, unlike the TurboTap that required human intervention to make it come into existence through practical exploration of Nature's unmapped possibilities.

This fundamental distinction between software patents and the other kinds is reflected in all the problems that are cited with the former: the fact that they are patents on knowledge, and the fact that you often can't invent around such patents, because that's like trying to invent around logic.

Most commentary has concentrated on the claims about software patents, but there is another that I think needs rebutting, since at its heart lies a profound misapprehension about patents today.

Here's the key paragraph:

Now, you can argue about the length of the patent grant, and about what specific inventions should be granted patent protection — these are all important and ongoing arguments. But the fundamental basis of the patent system is full disclosure from the inventor in exchange for an explicitly limited term of protection, and any effort to identify problems and reform the system has to respect the value both sides derive from that exchange.

That's certainly true, but the question that needs to be asked is whether the benefit obtained from patents through such disclosure is now being outweighed by the cost to companies and society of the litigation over patents that the growing patent thickets are giving rise to.

As I've argued elsewhere, the key issue here is that the patent system was created in the 15th century, when inventors and inventions were scarce; disclosure was extremely valuable for the reasons the article rightly emphasises. Today we live in a world of inventive abundance: there is simply no shortage of inventors or inventions. So we no longer need to pay the price of granting intellectual monopolies to people. People will still invent and make money from their inventions even if they are not protected by patents. Because the fewer patents there are, the more valuable each becomes, which encourages more people to invent until equilibrium is attained.

Ironically, the article I've been exploring provides a good example of why the patent system is grinding to a halt, and why it is simply not sustainable.

In his discussion of disclosure, the author points to Apple:

all those Apple multitouch patents are more than just attempts to prevent competitors from using a specific technology — they’re also detailed instructions for building that exact same technology in the future. Here’s a part of US patent #7,812,828, which Apple’s particularly fond of asserting in lawsuits: it lays out a system for tracking multiple finger and hand inputs on a multitouch surface and correctly filtering them.

(Amusingly, the two equations that follow, presumably quoted to impress us with their mind-bending complexity and originality, turn out to be a formula of speed - distance divided by time - and basic Pythagoras. Both are important, but of course trivial from a mathematical viewpoint....)

The patent in question is for "Ellipse fitting for multi-touch surfaces". As is customary, it begins by listing all the other patents that it cites. By my rough count, there are over 250 such citations of relevant technology. Judging by the dates they were granted, most of them still seem to be in force.

Now, some of them belong to Apple, but most of them do not, as far as I can tell. Since they are cited, they presumably have some relevance to the current invention, at least in terms of forming the intellectual background against which it was devised. I wonder how many Apple has needed to licensed because of that. After all, if it cites them, presumably at least some potentially represent important inventions that Apple is building on directly. Moreover, the ability for patent holders to block others from using its invention in further inventions means that there only needs to be *one* patent that its owner refuses to licence, and Apple has a problem.

I don't know about the particular details here - it might be that the citations are sufficiently distant from Apple's patent that they are not an issue. But 250 citations is a big number, and the bigger this number gets in patent applications, the more likely that at least one of them will demand royalties or block the new patent. Indeed, we are already seeing just such problems in the area of smartphones, where the patent thickets are already hampering innovation, and raising prices for customers as a result.

It's this downside of patent abundance that is the problem today. But as I've suggested, patent abundance is also the solution, because it means we don't need to provide an incentive to invent stuff any more.

The main problem with the post discussed here is that it doesn't step back to look at the bigger picture. Although it rightly discusses the original rationale of patents, it fails to relate that to the very different circumstances surrounding inventing today. When you do that, you find that abolition really is just a question of doing the maths.

Follow me @glynmoody on Twitter or identi.ca, and on Google+


Jeremy Bennett said...

Hi Glyn,

Like the article, but I think you are straying into a couple of dangerous areas with a couple of your comments:

"... that is, we believe that we can, *in theory*, use maths to describe anything that exists..."

Yes, but as Kurt Goedel pointed out, if you can describe everything, then you must also have inconsistencies. So in a patent world where you want consistency, it is not appropriate to say "everything is maths".

"... the logic of a computer program is cast-iron ... If you simply apply the laws of logic and maths, you know exactly what will come out at the other end."

Unfortunately the Halting Problem shows this is not generally true. You cannot even show whether a program will always terminate, let alone what will always come out at the other end.

Your philosophical argument is spot on, but limits of maths and logic have to be recognized when considering the issue.

Glyn Moody said...

@Jeremy: thanks for the comments.

You're right, of course. I was trying not to get too deep given the subject of patents is dry enough as it is...

So my comment about maths being able to describe anything was actually more flippant than it seems - merely the claim of mathematician about his subject being able to tackle anything, not that it was a perfect description of everything.

And as for the Halting Problem, I did write: "Although there are deep issues of checking whether programs work" which was a nod in that direction without going too deep.

After all, when you're writing an Android app, I doubt whether it's really a big problem...

So, yes, what I wrote was not totally exact - maths and logic have limits - but I hope that it doesn't undermine the basic ideas (although I'd be interested in hearing an argument that it does...)

gregory said...

patents are only a protection racket ..

obviously created in the west, with its notorious addiction to individualist-ism and its deadly concept of being separate from nature

Crosbie Fitch said...

Where did this idea of encouraging disclosure ever arise except in needing a pretext for monopolies that were thoroughly discredited on every other basis?

Patent doesn't encourage disclosure of anything except that which would otherwise have been an embarrassment to disclose.

1) All useful inventions that can be exploited without disclosure won't be disclosed even with patents.
2) All useful inventions that can only be exploited if they are disclosed would have been disclosed without patents.
3) The only inventions that are left are the useless ones, that are only patented on the off-chance someone might try using them, and thus be a lucrative target for patent litigation.