RT @ev: "There is no positive evidence for

Strongly agree - LANjackal

Strongly agree with that. The only value they provide is a competitive edge, and even if they were allowed for that reason, they become increasingly less relevant and should have an extremely short lifespan 1-2 years max. If you haven't diversified you're product in that time, you aren't innovating and others should be allowed to work with your basic ideas anyways. - OCoG of FF, Jimminy

Netflix successfully used its patents against copy-cat products for more than 1-2 years. Most of the cease-and-desist letters did not end up in litigation. - Piaw Na

Piaw: and that improved or increased innovation? - Bret Taylor

I'd argue that the Netflix prize would not have happened if Netflix had to spend all its resources defending against copy-cat products instead. - Piaw Na

And seriously, Google's been around for 10 years, and hasn't diversified its revenue stream to a substantial degree yet. I don't see diversification of revenues as being a given. Consider Microsoft's attempts to diversify its revenue streams. Much of those attempts to diversify were abusive, and the rest were lack-luster at best. After 20+ years at it, Microsoft is still largely about OS + Office. - Piaw Na

I really don't follow your line of reasoning. I don't see DVD distribution or video distribution as a bastion of innovation, even as a Netflix customer. I also am not sure why the Netflix prize is relevant or how diversification of revenue is related to this topic. It seems pretty much the same as Amazon's "one click" patent pursuits in the nacent stages of that industry; it is not clear to me it was actually necessary for Amazon's eventual success or that it helped "innovation" in the industry in any way. - Bret Taylor

Piaw, I never said anything of diversifying revenue, I said diversifying your product away from the original patent. - OCoG of FF, Jimminy

I think the Netflix prize did drive important work in recommendation engines. Whether you consider that important is irrelevant. Netflix innovated in proving that a subscription model can work for rentals. Not all innovation resides in code. - Piaw Na

Jimminy, and your stance then is that for Google to keep its ranking algorithms secret (under trade secret law) rather than publishing them (either in the form of patents or papers) is a good thing, right? - Piaw Na

A lot of people view patents from the point of view of "they're evil, they're bad." But forget that the purpose of the patent system is to get people to publish. I think it's extremely important. The forthcoming gtags patents for instance, describe important improvements in code search ranking. They may seem obvious, but I can assure you after watching another internal project spend 6X the man-power we did in our 20% time and still fail to achieve better results, it wasn't obvious when we built it! Keeping this stuff secret under trade-secret law is probably not in society's best interest. - Piaw Na

What if a patent could be used to stop spammy technology? That is where I see the positive in patents. - Jesse Stay

And don't forget, Xerox's huge revenue from holding the photocopy patent allowed it to fund Xerox PARC, which was responsible for much of the shape of computing we hold today. AT&T's long-time monopoly also allowed it to fuel basic research into things like the transistor. I'm not saying that the patent system can't be improved, but the "patents-are-bad" crowd aren't looking deep enough into the patent system to see what problems it's supposed to solve. - Piaw Na

Piaw, I never said that patents and papers shouldn't be allowed. I was saying that the life of such patents shouldn't last long. The ability for a company to prosecute you for using their publicly released data, in a hyper-paced environment is stifling to innovation. - OCoG of FF, Jimminy

Piaw: first, this comment is about software patents (unlike the Xerox machine). Most people who are against them feel patenting software amounts to patenting ideas (with a lot of legal nonsense to make them appear as something more than ideas). All of your arguments make the assumption that their patents led to some success and innovation, and I am really not sure you have made your case other than saying "Company X had software patent Y, and later they did something good." - Bret Taylor

OK. Pure Software held the OCI (Object Code Insertion) patent, which became the basis for valgrind. Would this work have become widely available if it had been held under trade-secret law instead? Or would it simply have died when the business concern died? Or do you not consider that innovation as well? - Piaw Na

What about PNG? Created to avoid the patent issues with GIF, it's far superior to it. Or GNU as an alternative to Unix? Patents force people to come up with alternatives that, at least some times, do lead to innovations in the field in which the patent resides. It's very hard to defend a universal statement like this. - Mark Trapp

Also, the ability to patent naturally occurring genes has not accelerated evolution. - Kevin Fox

But it *has* accelerated terrible movies/TV shows about genetics companies creating super heroes. - Mark Trapp

Gary's exactly right. Patent law was intended to justify expenditure by guaranteeing a period of exclusivity. GIF is a backwards case because Unisis didn't even attempt to monetize the patent until the exclusivity had nearly run out. PNG or something like it (probably an augmentation of JPEG2000) would have come to pass even without the GIF patent. The need for 8-bit alpha transparency grew concurrently with the GIF patent issues, but they were unrelated. - Kevin Fox

I disagree. Patent and copyright law is there to protect and promote innovation: forcing others to come up with novel and better ways to do things is innovation simpliciter. Had there not be a patent issue with GIF, people would've attempted to modify a broken format to begin with instead of starting over with a clean slate that produced a far superior result. I don't buy that we'd be in a better place had GIF been allowed to be used freely. Innovation rarely occurs unless a person's hand is forced to produce something completely new. - Mark Trapp

Kevin, JPEG 2000 vs. PNG is a great example of what I'm talking about. PNG comes about in 1995 due to finding out there's a patent issue with LZW, a full 8 years after GIF is introduced. Nowadays, everything supports PNG because there was a push to avoid using GIF due to licensing issues. JPEG 2000, which did not have this issue, is still not widely supported 9 years after it's introduced. There's no pressing need for a new format since JPEG is still acceptable. - Mark Trapp

Gary, not necessary, but PNG's development was spurred specifically because of the LZW patent issue: http://www.libpng.org/pub... Routing patents spurs innovation, whether intended or not. I'd keep patents around just to get people to re-approach problems in new and better ways. - Mark Trapp

Gary, I don't discount that possibility, but that's different than saying there's no positive evidence for software patents improving or increasing innovation in software. The reality is much more gray than that. I do wonder what the exemplars are where no innovation can occur due to a patent. The ones I can think of right now seem like a matter of maturity, not impossibility. - Mark Trapp

For every 1 worthwhile software patents you can find, there are 10,000 bullshit patents. I'm surprised Piaw that you don't recognize how busted the patent system is, how easily it is gamed, and how much of a chill this casts over the industry. The harm that has been done is IMHO far worse than the good. People are patenting anything and everything these days, every useless and obvious extention, effectively creating intellectual minefields everywhere. As for Netflix, I'd argue that businesses adapt to the environment they're in and that businesses don't just lose just because others can copy them. Patents are not needed to execute well and build brand loyalty. - Ray Cromwell

Besides, is 20 years in todays age really a legitimate period of protection of software algorithms? Pharmaceuticals I can buy, given the long time to develop and test them, but people patenting auctions, shopping carts, and 1-click are not spending $100 million on FDA trials and 10 years searching through idea space. I like Netflix, but nothing they patented wouldn't have been figured out, and while the Netflix prize is cool, they are not the only people vesting tons of money in collaborative filtering or recommendation engines. - Ray Cromwell

I can see some argument for software patents, in very limited form. I think that long terms, overly broad grants and submarine patents make them a net liability at the moment. - Nick Lothian

To be more specific: 2-5 year limits and use-it-or-lose-it restrictions should apply - Nick Lothian

@Mark, I'd say video compression patents have really gummed up innovation, since most of the basic concepts are already covered. The x264 community has shown that there are lots of refinements that could be made to H264 for example, but such an enhanced algorithm would be encumbered. It's not clear to me that Theora is really unencumbered, just unchallenged due to lack of use. Does Wang's "Save As" patent really help? What about Eolas's harassment of Microsoft so that now users are forced to click twice instead of once? In 1997 I "invented" CAPTCHA and filed a patent for it, I later sold the provisional patent to another company (which imploded), had the patent not died, pretty much everyone would be sued today who uses CAPTCHA, would that really be helping internet security or users, or would it just be lining the pockets of IP astroturfing law firms that produce nothing but lawsuits. - Ray Cromwell

Software patents brake innovation, and ROB Developing Countries where guys can't buy and defend patents. Period! - Thierry R. Andriamirado

If you are concerned about software patents be very concerned about Nathan Myhrvold, former Microsoft CTO, and his "Intellectual Ventures". He appears to be taking patent trolling to large scale. They don't actually file patent infringements, but may be farming the legal actions out to hundreds of shell companies, so they can deny they are a giant patent troll, but that may be what they are: http://techdirt.com/article... - Ed Millard

Mark, jpeg2000 adoption was slowed largely due to software patent fears. But let's suppose software patents do spur people to reinvent the wheel rather than spend time on totally new things. A slightly better wheel is likely less progress than we would have had if people's creativity wasn't forced into working around patents. - Bruce Lewis

Agree 1000 percent. - Todd Hoff

I don't believe that Software Patents are an unmitigated good. However, I also see lots of cases in industry where something wasn't published because it was a trade secret. The story I heard in particular was the Fast Fourier Transform, which wasn't patentable, but was kept a trade secret for a good 10 years before an academic rediscovered it and published. Google, Yahoo, and Microsoft all hold equivalent trade secrets today. I would rather get them to publish than wait for some academic to rediscover those ideas. - Piaw Na

I would posit that for every case where something completely new and better was invented in order to eschew a patented solution there are countless iterative and evolutionary improvements that were never realized because that innovation built on earlier, patented innovations. - Kevin Fox

Piaw, there's a simple way to bolster that argument: You write software, right? Just tell us about something you were able to do thanks to something you found in the patent database. That's what this is about, right? Promoting the art by publishing inventions. Are you making use of the published inventions? I'm not, as far as I know. I've been programming full time for 20+ years and never met anybody who approached new problems by searching the patent database to see if someone had already done it. - Bruce Lewis

"There is no positive evidence for software patents improving or increasing innovation in software." - but the purpose of patents is not to promote innovation - it's to give the inventor time to profit from his/her invention. I'm not defending software patents - I think the patent process is a bit out of control - but @ev starts with a phony premise. - John Craft

John, what jurisdiction are you talking about? In the US, the purpose most definitely is to promote the progress of science and useful arts. See US Constitution Article I Section 8. - Bruce Lewis

"In the US, the purpose most definitely is to promote the progress of science and useful arts." - yes, by giving the inventor a limited-time exclusive right to exploit the invention. Or, put another way, I don't think most inventors are motivated solely by, and only by, the betterment of mankind. (And my comments aren't limited to software patents.) - John Craft

The exclusive right is the means. Progress is the end. When the means shows results contrary to its end, the means should be changed. - Bruce Lewis

I'll buy that formulation. And the conclusion. - John Craft

Change is difficult when so many profit from the current formulation. - Todd Hoff

John, the patent system *is* intended to foster innovation by encouraging people to put more time and money into invention knowing that they'll have a period of exclusivity to recoup their investment. The premise Ev states is completely sound. - Kevin Fox

Kevin, instead of questioning the premise I should probably have said was that I don't think software should ever have fallen under patent law - there should have been a separate regime set up and enforced so that concepts couldn't be co-opted by patent trolls. Eli Whitney didn't patent the idea of the cotton gin, and then sue anyone who tried to build one - he patented a working device. - John Craft

Bruce: I'm able to do search and get useful results everyday thanks to the pagerank patent. Very useful, that one. :-) - Piaw Na

Piaw, that algorithm has been diversified away from the patent, it might be a key source to how they rank pages. However, I also know that it has stifled my own innovation, a few years ago I was working on a Economic Analysis of Value Transference through Markets, prior to reading the Page Rank paper. A few weeks later I came across the paper and stopped working because my independent work would have been seen as derivative, and ultimately culpable to a lawsuit because it would rely on computers to perform the calculations necessary. - OCoG of FF, Jimminy

By "make use of" I mean innovate based on the patented invention. Either you licensed it, or the patent expired. - Bruce Lewis

John: It's a slippery slope. He didn't patent a physical object; He patented a design. At what point does a design become an algorithm or a theory? - Kevin Fox

A design of a working device, though, not a very loose "somehow this thing will separate out cotton fibers" claim that undeservedly achieved patent protection. - Andrew C (✔)

Kevin, aren't algorithms just machines for carrying out a computation and thus should be patentable? Isn't a design just a machine for producing algorithms? And so on... - Todd Hoff

Jimminy: you can patent a derivation of an existing patent. Bruce: the pagerank patent sparked a lot of other innovation. But most of the follow-ons did not get published as patents. They are/were held as trade secrets instead. And I don't view that as a good thing! - Piaw Na

Andrew: I've yet to see a patent, software or otherwise, that tries to claim, in effect, 'somehow this thing will do that thing'. - Kevin Fox

Piaw, aren't patents supposed to provide an incentive not to hide inventions as trade secrets? If most follow-ons remained trade secrets, doesn't that show that the patent system isn't working in this area? - Bruce Lewis

Todd: for a good analysis of why algorithms should not be patentable, see Donald Knuth's letter to the patent office a couple decades ago: http://www.pluto.it/files.... "To a computer scientist, this makes no sense, because every algorithm is as mathematical as anything could be. An algorithm is an abstract concept unrelated to physical laws of the universe." - Bret Taylor

Piaw, I had a similar idea for coderank based on pagerank like you did years ago, since so many open source projects allows one to construct large dependency trees and callgraphs from software (I can tell exactly how many people 'link' to the Apache implementation of SHA-1 vs alternatives based on imports or callsites), inspired by GiantJavaTree and Koders.com, but I thought it would be inviting a lawsuit to do it. I think many people would love to apply PageRank techniques elsewhere (ImageRank being another straightforward application), but even if you end up with your own patent, people are still liable for the derivatives. - Ray Cromwell

BTW, isn't there an argument that if Google completely published everything they're doing, even if granted total legal protection, it would harm search results by allowing people to game the system? To some extend, trade secrecy helps anti-spam SEO. If Bing isn't licensing PageRank, and non-googlers aren't using its techniques, then in what sense did it's publication benefit innovation anyway? - Ray Cromwell

Bret I understand that line of thought, but I think it's less accurate than the machine model. I'm going for a reductio ad absurdum angle here. Algorithms are really the opposite of abstract. They concertize an idea, they make it directly executable in the same way a prototype concretizes an inventors idea for a device. - Todd Hoff

Piaw, it might be possible to patent derivative work, but it would still be hard based on any prior arts ruling. One plus, I see to the patent of PageRank, is that Google has no right to go after competitors for using it, Stanford possesses the patent, though they have provided Google exclusive rights to use it. - OCoG of FF, Jimminy

I think exclusive license means that Google does have the right to go after anyone else using. :-) - Piaw Na

In the case of PKP/RSA, the RSA public key cryptosystem could be said to be an algorithm, but it was essentially a mathematical formula, and could not be patented outside of the US, and even within the US, it was beaten in court. PKP practically tried to own even the concept of assymetric cryptography, one suspects to the delight of the US Govt/NSA which wanted to limit the spread of cheap, secure encryption schemes. Clearly a case where the patents were used to quash public benefit, not help it. - Ray Cromwell

MPEG-LA responds: http://lwn.net/Article... H264 is covered by over 1,000 patents. What is the probability that Ogg Theora nor VP8 is covered by any of them? This is an example of the minefield effect. How could anyone independently invent a new codec AND have the time or resources to make sure it if violates none of a thousand different patents? - Ray Cromwell

As with everything in life it is all about balance and moderation, most everything has negative effects in their extremes. For example, I think issuing a patent to Facebook for the newsfeed is just stupid, that is like the NY Times trying to patent the newspaper. Now, if Facebook developed a special algorithm to make what is in your feed more relevant I would say that would deserve a patent as it would be rediculous to think that anyone would develope these things if you had to give it to anyone who wanted it after you invested the time, effort, resources, risk & persperation in your project. Facebook's Newsfeed is nothing more than an adaption or an improvement on the RSS concept which they did not invent to begin with anyways. If I improve Google's search algorithm and get a patent should I be able to prevent Google from using theirs anymore? You can take support for patenting anything and everything to far and at the same time you can take opposition to patenting anything and everything to far. After all, these opposites are very similar when you think about it, both could lead to chaos. - Brian

Friends

Login

Service