Patents are living up to their ideals | Hacker News Reader

Patents are living up to their ideals

justinsb.posterous.com

18 pointsamirnathoo14 comments

One particular example doesn't make for a very compelling argument. The vast majority of software patents (probably above 99% of them) are A) never followed up with scientific papers, B) represent nothing that someone else trying to solve the same problem wouldn't come up with, and C) are written so incomprehensibly that no one could ever derive any intellectual value from reading them.

Throwing out counter-examples of "good" software patents doesn't really attack the thrust of the main argument put forward by the anti-software-patent people, which is that almost all software patents are total trash that should never be granted.

I think you can make one of foudarguments here. 1) You can dispute the assertion that almost all software patents granted should have failed the obviousness/novelty test and/or be invalidated by prior art. 2) You can argue that the patent system should be tightened up so that most patents aren't granted, but to still allow certain types of patents (encryption algorithms, whatever else) and then discuss how to classify what sort of patents really are useful. 3) You can admit that but argue that the good done by the "good" patents outweighs the evil done by the "bad" patents. 4) You can argue that you should ditch all software patents entirely because the harm caused by bad patents vastly outweighs the good done by good patents, and that it's too difficult to legally distinguish between the two, so on balance it's better to just disallow them

Again, keep in mind that copyright and trade secret law cover already cover the theft of the software itself, and that plenty of people online talk about what they do and how they do it without any sort of patent protection.

So I'm not sure which argument this article is trying to make (though it's obviously not #4): are you arguing that software patents aren't, on average, total garbage? That standards should be tightened up to allow only a limited subset of what's currently patentable, and that the Google patents fall under that umbrella? Or that the good done by these few patents somehow outweighs all the harm done by all the garbage patents and associated trolling?

It's totally valid to point out that not all patents are a joke, and that some good can come of them, but that in and of itself doesn't really address the real meat of the debate.

I'm arguing simply that Google's papers were IMHO the most influential Computer Science papers of the past decade, and they wouldn't necessarily have been published without patents. The patent system is designed to encourage exactly the sort of disclosure we saw here, in this particularly important case.

I chose this 'one particular example' because it's the inventions where the papers are published that are important; not the 'bad' patents which should never have been granted and where the USPTO overturns the patent on re-examination (as I suspect many bad patents, like the linked-list patent, would be.)

I like the idea of a standardized patent argument list! I'll take argument #3 please.

I'm a fan of clearly-delineated lines of argumentation since it's difficult to have a debate if people just talk past each other. It's much more fruitful if you can really pinpoint exactly where it is that you disagree with the other person: if I think "B" is true and you disagree, it's much more useful for me to say, "Well, I think A->B and A, so thus B is true" and then you can say "Well I agree that A->B but think A is false" or "Well I agree A is true, but I don't agree that A->B", and then we can have a much more useful conversation instead of talking past each other.

In the case of the patent issue, there are so many entangled issues that it's often difficult to find the real points of agreement and disagreement.

I'd be interested to see someone actually try to make the argument that more good comes from the current system of software patents than would come from a system with no patents, since from my perspective the balance is pretty negative. On the good side you have inventions/work that wouldn't have happened without patents, the resulting utility created by them both in the form of direct usefulness to end-users and in terms of economic activity they fostered, and the knowledge sharing enabled purely by patents and all its resulting downstream effects.

On the bad side, you have all the inventions and work lost due to patents: all the work not done because of fear of patent litigation, jobs lost due to companies sued of existence by patent trolls, all the resulting loss of utility to end users, and the economic cost of diverting so many resources to patent litigation.

Personally, I would argue that far more is not done or is lost in the software world because of patents than is gained because of them; if you just eliminated all software patents I think you'd see a net gain in worldwide utility rather than a net loss. Copyright and trade secret protections mean that most software activity would still happen (that's what pretty much everyone uses for protection anyway: only big shops have the resources to really use their patent portfolio, and then it's often purely defensive), so the main loss would be in terms of disclosure. What you'd lose in terms of, say, not reading papers on PageRank, you'd more than gain back in terms of startups that don't get shut down and by diverting all that money spent on the legal system to better purposes.

I agree that reasoned and clear debate is the best approach. I've tried to make the case 'pro' by citing the impact of PageRank/MapReduce/GFS, and suggesting that the impact is no less than reshaping the way we think about computing.

So what are the (non-anecdotal) examples of the costs? It's also important to put those costs into context, because everything has negative unintended consequences in the real world. What's the total annual expenditure on patent litigation benchmarked against (for example) medical malpractice cases? Which companies have been shut down by patent trolls benchmarked against (for example) the companies shut down by new environmental legislation? Which inventions are available on some platforms and not others (e.g. ClearType), and what's the productivity cost of them benchmarked against (say) not pursuing remedies against the Microsoft monopoly?

See this book for a complete discussion on these topics and more: http://levine.sscnet.ucla.edu/papers/imbookfinalall.pdf

That book weighs all of the costs and perceived benefits of patents form every angle in every industry and makes a very fascinating conclusion.

The "disclosure" argument would hold more water if people actually read software patents, or went out of their way to license patents from other companies. But it seems to me that the vast majority of software patent cases are ones where the infringing company has either come up with the invention independently or imitated the external behavior of something else. In which case, the disclosure aspect of software patents is practically worthless.

The article argues that we wouldn't know how Google works without patents, but the truth is that those Google patents alone don't get you Google performance -- most of their system is protected not by patents but trade secrets. And this has not stopped Yahoo, Bing, or others from coming up with viable search engines.

But people do read the papers, and I'd argue that the papers are only published because the patent has been filed. I'm willing to bet that 80% of the hackers here have read the MapReduce paper (and for anyone that hasn't, you should!), but I agree I'd be surprised if people have read the patent.

Independent invention is a problem, and that's why there's a non-obviousness test for patentability. Of course, that's a very tricky thing to implement in the real world. So what you're really arguing is a higher standard of non-obviousness; I hope you'll be submitting that proposal to the USPTO.

I'd actually argue that Google's performance _does_ come from MapReduce. The trade secrets (that we know of) are around their customized hardware and the weighting factors in their algorithm, but I'd argue that makes a relatively small difference compared to Google's big idea, which was figuring out how to harness together lots of cheap computers, rather than investing in making each computer more efficient (e.g. Altavista, whose original raison-d'etre was to demonstrate how the DEC Alpha computer was powerful enough to support searching the whole Internet of the day)

It would seem to me that I cannot use the described algorithm and put it to work without landing in possible legal troubles.

These patents may actually slow down technological development rather than increase the pace.

We get the fringe benefit of knowing something for real but we are unable to put it to work.

In the world where open source hardware companies like Makerbot Industries are coming into existence and that the phrase "the atoms becoming the new bit" is becoming more true everyday, this is a worrying obstacle in technological progress. People cannot voluntary share and benefit from each other without a fear of patent holder infringing on their ability to do so, no matter how innovative he might be.

The patent system is incompatible with certain business models that relies on the openness of their product, and a whole path of technological development.

This is the price we pay for assuming that people won't share information about algorithms and other useful knowledge. The only question is, will the positive black swans of the patent world will truly outweigh all the lost potential of the open source paradigm?

Just remember, positive black swan in the open source world can happen too!(Linux, Mozilla Firefox, and more)

Academics continue to publish papers in fields where patent protection is unavailable (history, literature, etc.), so I fail to buy the argument that no scientific papers about computing would be published if it weren't for software patents.

The author uses the timing of Google's patent filings and published papers to imply that Google would not have published the papers without patent protection. I think the timing of the events does not provide enough information to determine what Google would have done without the patents. Google might have freely published the papers in a world with no software patents.

I'm the author... and yes, the 'implication' is exactly what I'm saying. As I said in the post, you can argue that Google would have published in a world without software patents, but the purpose of patents is to incentivize publication where otherwise advances would remain secret, and that's the sequence that we saw. In that sense, we're seeing exactly what the patent system envisages.

Your point of view seems to be that companies would publish their secret sauce freely without IP protection, for the warm fuzzies / indirect benefits that would result. You're free to believe that, but the patent system is based around the idea that corporations don't behave so altruistically - that people with billion dollar ideas don't want to give them away.

It is probable that Google would have published their algorithms after a long period of time. By the time that happen, the old algorithm would probably have no competitive advantage anyway.

I once read/heard an argument against the patent system in that the patent system didn't really help disclose secret inventions.

What it does is shift the focus of inventions from the highly secretive kind to inventions whom inter-working would be rapidly discovered anyway.

Something to think about.

(BTW, everyone should read Against Intellectual Monopoly at http://levine.sscnet.ucla.edu/general/intellectual/againstfi...)

Google only published because they felt secure with the advantage conferred by their patents?

That idea only makes sense if Google now starts enforcing their patent-granted monopoly on these techniques: suing people who are doing the things they've published. They haven't.

Until they do, this model of Google's patenting/publishing motivations is flawed.

Reply on news.ycombinator.com