Daniel Wallace

Subscribe to Daniel Wallace: eMailAlertsEmail Alerts
Get Daniel Wallace: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


LINUX VIEWPOINT: "GPL - Are We Wishing for Something We Really Don't Want?"

LINUX VIEWPOINT: "GPL - Are We Wishing for Something We Really Don't Want?"

  • Read Daniel Wallace's original Letter to the Editor
  • Read response to Daniel Wallace by Celia Santander Esq.

    This is neither a legal conclusion nor a philosophical statement of bias but merely conjecture for thought.

    Suppose all the programs ever released under the GPL are placed in a non-commercial public domain trust fund.

    There are a few million lines of copyrighted code in a Linux distribution. Add all the code at SourceForge. How many lines now? 100 million?

    How about everything out there? Technology is growing exponentially. How many lines in ten years? 10 billion?

    How many in 20 years? 100 billion?

    One requirement to show infringement is access to the code. If it's publicly available then that element of infringement is a foregone conclusion.

    What does a commercially motivated original author do to check that the ever larger growing pool of code structures that exist out there in non-commercial GPL trust are not "substantially similar" to to his code? Is he infringing?

    How long does it take to check with ESR's code comparator program? ESR says 2 million lines per minute on a 1.8 GHZ Intel Box. The implementation is O(n(log n)).


    (100E9)log(100E9) = 25.3 x 100E9.

    So 2.53E12 / 2E6 = # of minutes = 1.26 million minutes. (check the math)

    That's about 2.4 years to find out if you have any substantially similar code that infringes on the exponentially growing pool of GPL'd code. Of course the GPL code pool grew during that 2.4 years.

    Is that practical or desirable? Are we wishing for something we really don't want?

  • More Stories By Daniel Wallace

    Danniel Wallace is a retired physicist who has been connected with Open/Free software
    for some time. He is an associate member of the FSF

    Comments (17) View Comments

    Share your thoughts on this story.

    Add your comment
    You must be signed in to add a comment. Sign-in | Register

    In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.

    Most Recent Comments
    Re: Bill P. 02/10/04 07:15:17 PM EST

    I think you misunderstand what GPL really is. GPL is a license under which you distribute your COPYRIGHTED software. It is NOT public domain software, and you retain your copyrights and the right to distribute it under a proprietary license. Shareware does not serve any function other than to try to get people to try your software, and it produces poor quality buggy software - it is not a shared development method and doesn't save development costs. The latter is why GPL software is taking the software development world by storm.

    One way of making money from GPL is by GPLing software and then offering an alternative proprietary license to proprietary developers. The GPL license ensures the software is established as a standard, gets bebugged and maintained for minimal cost, and more importantly, prevents other people from ripping it off and selling it as a proprietary product because they can only sell their product as GPL unless they license your code from you under a proprietary license. This allows you to license out your code as a proprietary product for those who want to link it to a proprietary product. Trolltech does this with it's QT libraries, Borland does this with it's Interbase SQL server, and so do a lot of others.

    Another take on this - Why does nobody sell cheap SuSE Linux CDs? Simple. SuSE sells it's installation and configuration program YAST under an open source license that allows anybody to modify it if the modification is made available for SuSE, and copy and redistribute it freely provided it is redistributed free as in $0. The result is no cheap SuSE CDs and everyone paying SuSE about $90 for each CD set. Why? because nobody is going to burn a CD which costs them 50 cents and give it away free. GPLing something works in a similar way. It protects the copyright holder's source code which has been made readable by the public, by forcing those who might steal it to give the source code free or at a very low price. This is why GPL software is so popular with software developers.

    Bill P. 02/10/04 05:45:58 PM EST

    I think for the open source and GPL to really work the computer industry will have to adopt new economic rules and get away from protecting every line of code. There are benifts beyond financial gain.

    There are ways to share your programs and the code without giving up your copyrights.(or money) for expample shareware which has been areound for years. In the early days of PCs code and programs were share freely, the idea was that other can learn and benifit from them. The is the spirit of open source and GPL.

    The way I understand the GPL and open source is that everyone can benifit from the work of others, be it learning mew and/or improved programming skills, inspiration to build and/or improve on the work of others or even greate something totally new.

    If you are looking to make money from your program(s) or code then you SHOULD copyright it and protect it from prying eyes. But that is'nt what open source and GPL is all about.

    Just my humble opinion.

    eric seynaeve 02/10/04 08:59:59 AM EST

    I'm sorry for the 'scientist', but his science is flawed. If his assesments were true, Google would never exist (it searches over 3 billion PAGES in less than 1 second).

    Assuming that one would use ESR's algorithm to search a public and known codebase versus his own know codebase is silly. ESR's algorithm is about a different problem set (search one codebase versus another codebase without revealing the underlying code itself).

    As any good carpenter knows: use the correct tools!

    NoName 02/10/04 07:35:17 AM EST

    It is amazing how much crap you can read theses days from scientists (being one myself...).

    Why should I bother to check all GPLed code, if I am writing new software? The only reason I can imagine is if, and only if I want to write GPLed software and do not want to reeinvent the wheel.

    It is just not true that I have to watch out for a similar piece of software, because my work is neither based nor derived from the GPLed code. If you are afraid that you might accidentally write code that is so similar that is might look like being derived, then you are in big trouble anyway. Or would you like someone to claim that your product is based on public domain software? Or being sued by a company for "stealing" their code (which you have never seen before?) By the way, chances are much bigger that you are sued before for violating a software patent, the plaque of modern IT industry.

    Please have a quick chat with your lawyer, if he is a decent one he might be able to explain the difference between patents, licenses, and copyrights (while you are at it, just ask for trademarks as well).

    The whole matter is sufficiently confusing to excuse a scientist for calling left right and right left, but please do not embarrass yourself again by making your lack of knowledge public.

    Thank you

    Anonymous 02/10/04 05:05:45 AM EST

    This is a red herring, the author confuses copyright and patents. Copyright does not prohibit independent creation of something similar by chance. Independent creation will be plausible mostly for stuff without sufficient creative value, and that is not protected by copyright, anyway.

    Wevervain 02/10/04 04:44:59 AM EST

    I also read your last letter. That was bollocks as well.

    anonymous 02/10/04 03:55:32 AM EST

    You say in future it may take two years to check if your code infringes with GPL code. How long do you think it takes and costs now with proprietary code? The other company has to file a lawsuit and go through a discovery process to see your code to check if you have infringed. This takes 5 years or more and both of you have to pay huge legal fees.

    GPL code is much much faster and cheaper to check, and is a lot less litigious and developer friendly. If you do infringe, Instead of launching a SCO type extort-as-much-as-you-can lawsuit, the FOSS will ask you politely to either remove the GPL copyrighted code in your software, or to GPL it. The choice is yours. This is a lot better in every way than what you have currently with proprietary software.

    Andres Garcia 02/10/04 03:17:12 AM EST

    How can an original software author be certain that his code doesn't infringe on GPL code?

    Quite easily, by writing it himself.

    Simple solutions are often the best.

    Philip Stephens 02/10/04 02:56:01 AM EST

    Who really cares how much GPL'd code we end up with? If it truly were to become a problem for proprietary software makers, this just demonstrates the failings of the current copyright system. Remember: RMS just wanted his code to be free, and he used the existing copyright system to achieve that goal. If the GPL ends up undermining the copyright system, that's the fault of the system.

    Jean-Marc 02/10/04 12:02:20 AM EST

    OK, so now source code grows exponentially? I didn't know people are getting exponentially more productive. Cool! Even if it was important to check against all GPL source code that exists (it's not), it should be noted that processor speed grows exponentially while source code probably grows linearly (merely because it accumulates). This would mean it would become easier to search over time. Last factual error, if you're going to search in a really large database, the complexity can be made O(N) using hash tables and similar tricks. Then again, you don't need to search if you haven't copied anything.

    AnonProgrammer 02/09/04 11:04:28 PM EST

    No said that you have to go with the GPL to be open source or that you code had to be open in the first place.

    The GPL only says that if you modify a program that is licence under the GPL that your changes are also licenced under the GPL.

    Matt 02/09/04 10:31:26 PM EST

    Ahh, so let me get this straight, authors who are giving something away for free, are going to hire a team of lawyers to search day and night for your commercial project's source code to see if it's infringing? First off, most GPL'd code will not be patented, so the algs etc can be used. The whole spirit of GPL is sharing, and only in fairly obvious circumstances have they gone after anyone. The logic in this article is completely flawed.

    Dan O'MAra 02/09/04 08:35:13 PM EST

    How long does it take for the author of publically available GPLed code to determine if he is infringing on unpublished "super-duper-secret" proprietary source code?

    Since we'll wait forever for a reply, here's an easier one.

    How many physicists does it take to screw in a lightbulb?

    Answer: none, the technician does it so the physicist won't get hurt.

    Congratulations Dan Wallace on your new found position as contributing author to LinuxWorld.

    BSDProtector 02/08/04 04:00:52 AM EST

    Mr Wallace is at it again. One misleading premise and we start thinking about big implications...

    True, access to code is a requirement for proving copyright infringement. But, in a real case the code would be traced all the way through an organization, from the download (if any) to the version control system etc. It would be really easy to figure all that out.

    Bottom line, if you didn't actually copy, you have nothing to worry about. What's more, it's not like GPL police is going around busting people for infringement.

    And to answer the question, yes, we do want all that code GPL-ed. The only problem is, GPL is an invalid license (according to Mr Wallace), so it's all for nothing anyway ;-)

    David Mohring 02/07/04 06:22:41 PM EST

    As for Daniel Wallace's theoretical potental good of putting existing GPL code into the public domain without any license. Check out this list of public protocols that Microsoft has embraced and extended, many to the point where the result is no long full interoperable...
    ... and put them under the Microsoft's Communications Protocol Program (MCPP), a license so unacceptable that the majority of even the proprietary software vendors refuse to have anything to do with it.

    Although Microsoft raises the issue of GPL violations, that is a classic red herring. Many more people find themselves in violation of Microsoft licenses, because Microsoft doesn't allow copying, modification, and redistribution as the GPL does. Microsoft license violations have resulted in civil suits and imprisonment. Accidental GPL violations are easily remedied, and rarely get to court.

    It's the share and share alike feature of the GPL that intimidates Microsoft, because it defeats their Embrace and Extend strategy. Microsoft tries to retain control of the market by taking the result of open projects and standards, and adding incompatible Microsoft-only features in closed-source. Adding an incompatible feature to a server, for example, then requires a similarly-incompatible client, which forces users to "upgrade". Microsoft uses this deliberate-incompatibility strategy to force its way through the marketplace. But if Microsoft were to attempt to "embrace and extend" GPL software, they would be required to make each incompatible "enhancement" public and available to its competitors. Thus, the GPL threatens the strategy that Microsoft uses to maintain its monopoly.

    Microsoft claims that Free Software fosters incompatible "code forking", but Microsoft is the real motor of incompatibility: they deliberately make new versions incompatible with old ones, to force users to purchase each upgrade. How many times have users had to upgrade Office because the Word file format changed? Microsoft claims that our software is insecure, but security experts say you shouldn't trust anything but Free Software for critical security functions. It is Microsoft's programs that are known for snooping on users, vulnerability to viruses, and the possibility of hidden "back doors".


    Chaz 02/07/04 06:01:55 PM EST

    Following that esr logic are you?

    A+B=C, so D must follow because esr says so...

    The author of the article is talking non sense in the same respect esr does when he declares things like the FSF will implode.

    DaveF 02/07/04 05:50:41 PM EST

    Whatever... I see LinuxWorld values controversy over accuracy, turning your contribution into a column in its own right. So much for journalistic integrity, I guess.

    In your letter to Ms O'Gara's mailbox, you made some ridiculous and false claims about the GPL. You ended your letter with this quote:

    "A quick check by a competent attorney with the citations above will quickly confirm the accuracy of these conclusions."

    When an eminent attorney responded to your "conclusions" and destroyed every one of them, you failed to even admit that you were wrong.

    Now, you're back here with this whole thing about "Suppose all the programs ever released under the GPL are placed in a non-commercial public domain trust fund." Yes, and what if angels didn't really have wings -- could they fly anyway?

    This is really becoming quite tiresome.