LinkedIn Twitter
Serial Entrepreneur, 5 startups, Product Guy, and Inveterate Blogger!

13 responses to “The Problem with Software Patents”

  1. Tom Donaldson

    Clearest explanation of when/where/why patent protection is applicable that I have ever run across.

  2. Steve Stites

    It is interesting that you used “Gone With The Wind” as an example of a copyrighted work. There is a parody of “Gone With The Wind” called “The Wind Done Gone”. Margaret Mitchell’s heirs sued the author of “The Wind Done Gone” for copyright infringement. You can read about it here:

    http://en.wikipedia.org/wiki/The_Wind_Done_Gone

    ————————–
    Steve Stites

  3. Dan Christian

    There is also this idea that someone must have copied an idea to come up with something similar. In software, it’s often impossible to see the original, but “copies” happen all the time. This is because the problem space defines the “innovation”, not the development process. Some argue that all mathematics is discovered, not invented; and that all software is really just mathematics.

    The patent office should retract patents as “obvious” when the same ideas keep re-appearing without any indication of copying. Or maybe the patent office should have to pay the legal fees when a patent is challenged and found to be obvious or overly broad.

  4. JB

    Another issue with the “protecting the little guy from the big, bad corporation” myth of patents is the sheer effort & expense to file a patent. $10-15K (the cost of a decent patent lawyer) might not be a lot to a big corporation, but it could easily be prohibitive to a small inventor, and there’re scads of people (including the USPTO) warning about the dangers of self-patenting. (After all, do you trust yourself to write a document that’ll stand up in court?) And, of course, the little guy would have to sue the big, bad corporation and its legions of lawyers if it actually infringed.

  5. Jose_X

    In disagreement with what may have been implied in part of the article, I would say that patent law is damaging in many circumstances (even besides software) and that copyright also goes overboard as well (eg, too long duration and “derivative works” interpretation borders on the degree of broadness found in a patent).

    I also would disagree with any possible implication that patenting should be more expensive (some readers might draw this inference after reading the article). This would just price out more inventors while allowing the multinationals and the very wealthy to still patent trivialities and own an even larger share of the IP pie. In fact, if we make patents much more accessible (and ideally implicit as is the case with copyright), then more people would have protection against attacks since fewer attackers would be able to pull off an attack successfully and without becoming vulnerable at the same time.

    So, while not endorsing everything the article above may have implied, I really agree with the main idea told through the 4 quadrants.

    BTW, it would be useful if this posting was under a CC license so that its detailed arguments can be harvested more easily into places like the wiki at http://endsoftpatents.org/ .

    Ref also:
    http://www.networkworld.com/community/node/60939?page=1 “A patent system that wrongs needs fixing”
    http://opensource.com/law/10/5/total-victory-patent-lawsuit-against-open-source-software#comment-1418 “Unconstitutional, illogical, damaging, hand-cuffing, insulting..”
    http://markproffitt.com/2010/04/20/details-of-abundance-based-intellectual-property-system/

    The third link just mentioned addresses the obvious: stop creating monopolies if our goal is to promote the progress of science and useful arts. It suggests we pay inventors through income taxes on everyone. The deal would be that everyone could use any invention for $0. There are other variations possible that are not described in that article, but the overall approach seems much saner than our current monopoly-based system.

    We should not tax software (eg, with patent monopolies) but should tax the greater output that would result from a more efficient and wealthy society able to access a larger amount of software.

  6. Mullington

    This has to be one of the more confused attempts at explaining patents. I think that I agree in spirit, but the text moves all over the place so I am not sure.

    1. Patents are intended to allow a limited form of monopoly, in exchange for which the inventor has to share the details of the patent.

    2. The patents granted shall not be overly broad, variations are allowed. From an inventor’s (or perhaps more so a patent troll’s) perspective a broad patent is good since that makes it more difficult for competitors to create non-infringing alternatives, but that is not generally the intent of the patent office. In regards to the example with screwdrivers I would think that the shape of the head is important to the patent, and you could very well patent that. Also, you could probably not patent a writing tool, but a pencil and a ballpoint pen are both different implementations and thus separately patentable. A different example concern pianos. I do not remember details but at some point in time piano makers used to patent their way of building a piano. Competing piano makers did not want to pay license fees so they had to make subtle variations – the end result was still a piano, but since the implementation was different enough they could still make and sell them.

    3. You are not supposed to be able to patent ideas. This is probably where the software patent problem lies since they are sufficiently close to being ideas. The patent system was created for physical items and the practice of allowing patents on less tangible things makes the whole rule set difficult to apply for the patent engineers at the PTO.

    4. Disclaimer: I am not a US person so I might have the wrong ideas about some of the things you discuss. Corrections appeciated.

  7. TJ

    Care to give us the reference to this “patented number” patent? I suspect you are mistaken in your understanding of what has been granted, but I’m ready to be proved wrong.

  8. Bob Warfield

    http://www.sciamdigital.com/index.cfm?fa=Products.ViewIssuePreview&ARTICLEID_CHAR=65B7FFE0-3A2F-4F29-93D8-6355A5F8F0A

    “Roger Schlafly has just succeeded in doing something no other mathematician has ever done: he has patented a number.”

  9. Jose_X

    Key parts of article condensed:

    Mullington, below are the main points I captured (and possibly filtered a bit through some of my own views). I agree with the argument being made in relative terms (software vs. machines, and copyrights vs. patents) but not in all details, eg, with the potential implication that copyright and patents as they exist today may be acceptable. [Copyright “derivative works” is too broad and copyright duration is way way too long. Meanwhile patents are way too long being such broad monopolies.]

    Software (literature, etc) should have copyright but should NOT have patents.

    Machines should have patents because copyrights aren’t enough.

    Software and copyrights:

    A software product involves creating and interfacing many components precisely. The digital (perfect) nature of software means that very complex systems can be created (mother nature/ physics does not impose very many limits). The high degree of allowed complexity mixed with human ambition and competition means a lot of intellectual investments exist throughout much of the almost always overly complex final product. This intellectual input scattered throughout the whole contribute to the value of the final product, ie, if we change enough details, we lose significant value, in part because the code that replaces it likely also had similar costs (eg, with debugging).

    We also have that it is very easy to copy (mimic) software to make more copies.

    The argument then is that something like copyright serves a purpose here because the original product has significant value throughout and which is very easy to mimic by competitors.

    [As an aside: The above description of software is much more true of source code than it is of binaries; hence, copyright is more important to protect source code. Almost ironically, the main proponents of open source (and not all shown source code is “open source”) frequently do away with much of the power of copyright. We note that open source’s value comes from the direct utility of the software source and from the derivable software binaries and not from the ability to bring in large income through license fees. In contrast, the biggest protection of closed source software is trade secret with copyright serving its use as well. Note that very successful software companies tend to leverage the hidden inner workings of software to maintain their revenue stream and frequently even allow strategic binaries to be given away.]

    Software and patents don’t mix:

    One reason to avoid patents for software is because patents are very broad by definition, and hence a single patent claim will necessary cover over many many distinct software products, each of which already would have involved a tremendous amount of hard work to conceive. We note that it is rather easy to write up a few lines of English to convey the claim the states in low detail the few properties that define infringement scope. Do we want a person writing up an overview of some properties to have command over a great many complex works?

    Machines and copyrights don’t mix:

    In contrast to this we have things like a physical invention. Here copyright fails for the following reason. The invention likely (a) has significant value and (b) represents a small percent of the final product.

    It has significant value in terms of high costs and time to discover. While, like for software, art, etc, it takes human intellect to define the final invention, there is an addition cost and time that are difficult to avoid in order to prod and poke mother nature into revealing its secrets. We can’t just conceive of something and then will it into reality as we can with software to some extent and with literature (think science fiction, for example). Generally, we have an idea of what might work, but ususally a lot of slow and more costly experimenting with scarce tangible materials is necessary. [I want to point out that this is much less true today as it was in the late 1700s because of the power of computers and our improved understanding of nature.]

    Copyright is subpar because the entire product will presumably have a few small components that hold most of the costly value (as described above). A competitor can change the nonessential parts significantly while retaining the key points and thus thwart copyright.

    Machines and patents:

    Patents avoid the problems of copyright in that a patent can protect the small essential portion specifically. Also, the product itself involves a lower degree of intellectual input (in contrast to software and literature which are pure products of the mind); hence, the broadness of patents does not lead to nearly as much other significant intellectual value by competitors being denied (or taxed) by the patent holder as would have been the case for software/literature.

    The article also points to a very important point: that we really do want to avoid granting monopolies. I wish the article would have then stated that current copyright and patents go way too far in how much monopoly power they grant.

    Finally, the article brought up at least one other important fact worthy of consideration when judging software patents:

    Patent law/practice is such that it is much more costly to prove innocence than it is to attack someone; hence, the attacker has an incentive to use a shotgun approach and fire off in the general direction of anything that moves. This will tend to “kill” many that could be innocent. Part of the reason for the cost disparity is that the costs to come up with the patent is low relative to the court costs required to annul the patent or prove innocence; the existence of the patent, as determined by the USPTO based on a very low hurdle implied by law (eg, which has no requirement at all of proving the patent would “promote the progress of science and useful arts”), shifts the costly burden of proof to the innocent. Further, the legal attack process can be optimized. Scales of efficiency apply since the attacker becomes intimately familiar with the attack process, the details of the patent, and the existing prior art. The person being attacked has much less experience with these details.

    In addition to the links I mentioned in an earlier comment, I’d like to recommend some reading on Nina Paley’s experiences with copyright:
    http://questioncopyright.org/sita_distribution_project
    http://blog.ninapaley.com/2009/12/07/i-2/

    and an interesting piece about what the framers of the Constitution may have meant when they wrote the “promote the progress” clause: http://www.digitalproductions.co.uk/index.php?id=177

  10. Jose_X

    >> Instead, I would argue that patents are plainly intended to increase incentive to publish.

    We have to wonder how a low detail patent can get monopolies while the vast quantities of practical, detailed, de-bugged and immediately useful open source gets nil. Something is definitely wrong.

    It would instead make sense to give open source innovation a full 20 years (rather than wimpy “prior art” status) or else lower the patent monopolies for all software down to zero as well.

    Goose/Gander

    >> and therefore not publically available to be used or improved upon.

    Unfortunately, when you grant a 20 year patent, the public has absolutely minimal incentive to attempt to improve upon the idea or use it (it can become illegal to do very much with the idea). For many years, the risks grow every single year you have added more investments to something that was patented since you can have a large part of your work nullified. And, worse, you likely didn’t know there was a patent or, if you had been told or got lucky researching, you probably found it difficult to avoid infringement because patents are so broad.

    Patents serve to rob people of their hard work. See this comment, for example: https://www.enterpriseirregulars.com/17600/the-problem-with-software-patents/#comment-7940 .

    Again, recognize and give monopolies to everyone (rather than simply to the wealthy) or else give it to no one.

    [BTW, no we can’t invent a better wheel than the round wheel, though even if we could (in some cases), the costs to society for everyone to try to go around the round wheel is a large one. There is quite a cost to a monopoly that people will discover in much less than X years or where they already have and there are few ways to go around that. Many things become clear to many people around the same time because there are in today’s world many with access to watch society well and to note certain things when the social context has reached the right point. Yet even if others are a few years our of the state of the art, why deny them the opportunity when they do get the chance? Point is that always someone if not a great many are being denied, especially for a monopoly that lasts way past the token year or two. Note that many many hard-working and smart people publish all the time (eg, mathematics, physics, open source, etc) without requiring a patent. The patent system, again, allows society’s more greedy and wealthy members to repackage largely others’ work in order to block society off from a very important class of solutions. If the patent holder got there fast enough, we all can suffer greatly. If someone beat the patent holder to the punch, the patent might still do much damage because of the costs to nullify it. And if a patent was not awarded, then society simply dodged one bullet among many coming our way.]

  11. Jose_X

    Sorry, the comment just posted (ie, “Goose/Gander”) was posted to the wrong website.

  12. sean

    I assume the Schlafly patent you refer to is this one.

    Claims of US 5373560 (A)

    I claim:

    1. In a cryptosystem, a method for inverting a cryptovariable comprising inputting said cryptovariable and representing it as an integer; constructing an approximate inverse to said integer; computing a residual of said inverse by taking a twos complement of a product of said integer with said inverse; testing a spill word included in said residual; improving the accuracy of said inverse until said spill word is zero; partial multiplying said inverse by said residual and adding a shift of said inverse; repeating until desired length and accuracy is obtained; and outputting said inverse.

    Even if the result of such a method might be a number the claim doesn’t relate to the number per se. The number is not an infringement of this claim.

  13. Why Software Patents are a Terrible Idea -or- Why you shouldn’t be allowed to patent a better virtual mouse trap. | A Software Guy