The Need for an Open Research License

Posted 31 Jan 2002 at 17:52 UTC by exa Share This

I'm a computer scientist. I'm very fond of free software as it helps me to use my machine to the fullest possible extent. It is also a bless when writing research code, free software will almost always cover the basics for you. A computer scientist can handle most of the issues that would be difficult to accomplish in a proprietary operating system reliably and rapidly.

It is a frequent extrapolation in the free software community to liken research to free software: all research software should be released as free software.

In this article, I argue that research code has very different requirements from free software (ie GNU/Linux systems) and propose that a license suited to the needs of a researcher should be crafted.

Now that my Msc. thesis is completed I wonder which parts of it I should release as free software. I need a license that will help me keep the research open. I wouldn't like my research being worked on behind closed doors.

In fact, that is what makes research different from commodity software. GPL is designed for generally useful software thus allows for private modifications of all sorts. That makes a lot of practical sense, as many people will need to modify the software for their own use.

Nevertheless, if I GPL a research code somebody will be able to 1) Extend my work in a journal paper, and not even talk about the code or the license. 2) Use my work as a service, say on the WWW without ever making the code freely available.

Moreover, my software is not generally useful software. It isn't an operating system, a command line shell, a compiler, or a web server. It is a very specific piece of code that can only be used and understood by experts. And I intend to keep it that way for a long time (say 3 years)

It is an orthogonal matter whether a work in computer science may have any practical use. As Dijkstra said, computer science is no more about computers than astronomy is about telescopes. Much research software does not even have a slight connection to the real world.

That said, researchers have unique requirements that we should consider. First of all, a researcher does a piece of work for recognition. Second, it's the results that matter not the software itself per se.

Third and most importantly, being able to maintain scientific practice is beneficial for both the scientist and science in general. Some scientists, like me, believe that is possible through making scientific process more open.

I'm cutting the article here, as my goal is to collect feedback from fellow hackers. What kind of a license would you deem appropriate for the requirements of a researcher who favors openness in science?


Bogus., posted 31 Jan 2002 at 18:33 UTC by egnor » (Journeyer)

Your software is only for "experts" (researchers) performing a specific task (computer science researchers) -- so is lots of free software. GNU/Linux itself was originally built by and for "experts" (Unix hackers). Free software is hardly limited to mass-market consumer software!

Your software doesn't have a "practical use" -- yes it does! It helps computer science researchers perform computer science research. That's a perfectly practical use, no more or less than GIMP or Apache or the Linux kernel or any other free software tool which is useful in some situations but not others.

Researchers do work for recognition; isn't this what drives many free software authors?

It's the results the matter, not the software per se -- but that's true of all software. Software is only useful insofar as it does something. In your case, what your software does is produce results that may be useful in computer science research. In the case of Apache, what the software does is serve Web pages. In both cases, the effectiveness of the software at its task is the key thing.

Private use, Web services, making changes and using the results without redistributing your changes -- this concern is shared by many free software authors, it's not limited to science. This topic is its own discussion.

Finally: Just what sort of license are you suggesting? Make a proposal already, you're the one who thinks something different is appropriate. What kind of license would I deem appropriate? The GPL, especially since you seem to want to prevent people using your work in a proprietary way.

GNU General Public License, posted 31 Jan 2002 at 19:27 UTC by grant » (Journeyer)

the GPL may be applied to research in every field, and is an effective measure against piracy and monopoly. until copyright and patent laws are sufficiently remedied it is a much better solution than the public domain for works meant to be shared.

neither points are correct :

(1) derivative works must mention the original copyright

(2) any services, public use, or other derivative works must release all changes as source code

both the LGPL and GPL are written as such, but the LGPL offers the ability to create derivatives (compile/link) with proprietary works, whereas the GPL restricts its use to other freely/openly licensed ones and requires that derivatives also be licensed using the GPL.

the GPL is meant to be free from ownership (but not attribution for creators and maintainers) whereas the LGPL is primarily intended as a means of competing more effectively with existing proprietary offerings. I don't think the LGPL would be applicable for publications other than software.

as opposed to the public domain, a GPL'd work is protected in the sense that as long as abuse is vigilantly thwarted it will never be able to be made less free from restriction : everyone owns it as much as anyone else (authors are duly attributed and recognized) and it can never be patented or monopolized.

Re: GPL, posted 31 Jan 2002 at 19:35 UTC by egnor » (Journeyer)

Are you saying that services and public use of modified GPL software must release the source code to the modifications? That's not my understanding; AIUI one must only offer derived work source when *distributing* any derived work (such as an executable binary). The program's output is not considered a derived work.

(Your wording is unclear; I'm not sure whether you're saying that your (1) and (2) are correct or incorrect. As far as I can tell both are incorrect.)

GPL, posted 31 Jan 2002 at 20:46 UTC by grant » (Journeyer)

in response to egnor : yes, services and public use of GPL'd works constitute distribution and/or a derivative of the work

distributed computing is not exempt from licensing restrictions, including the GPL : it's still publishing, in fact it's real-time distribution of the work

don't tell me that a publicly available service built with a GPL'd work is not a derivative of that work : it is

from the GPL, section 2 :

b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License.
the program's output may not be considered a derived work per se, but the hosted and published service most certainly is a derived and/or redistributed work

section 3 goes on to cover the requirement of releasing the source for any redistribution or derivative, thus proving my first point

obviously my points (1) and (2) are meant to be correct and logically true statements... as for the fact that derivative works must carry the original copyright : section 2 describing restrictions on derived works explicitly states that modifications must adhere to the terms of section 1 of the GPL
1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program.
section 2c is a little strange, but is meant to ensure that the copyright notice and licensing agreement are directly associated with the published work and are publicly available - perhaps this is something that needs to be strengthened for applications of GPL'd works in distributed systems

remember that every distributed application is a publication

Re: Bogus.., posted 31 Jan 2002 at 21:12 UTC by exa » (Master)

Hmmm.

First, there is another distinction between research software and GPL.

GPL is bound to a particular technology. It depends on the notions of "binary" and "source" for instance. Scientific experiments do not always come in the form of UNIX programs. The terms of the license are not general enough to be useful for researchers. (I leave the solution as an exercise to the reader ;)

Second, GPL does not warrant recognition. Any researcher who is releasing GPL'd software for recognition is probably not on the right track. Tell me the names of researchers who contributed to GPL'd code. Can you? I can't. Research is not exactly clowning on mailing lists, flame-war-fests or popularity contest; Linus may have rockstar status but he isn't a researcher. (At least he didn't gain his recognition from any scientific work)

A license similar to original BSD, with an "advertising clause", is much more suitable for research work. If I write a program, and if I am the principal researcher I demand that my name is written on the top, every time the program is run, every time one looks at Help|About, in every documentation. Period. This rightful demand is not satisfied unless the license has an explicit statement about "whom" the researcher is and "how" he should be credited along with measures to prevent abuse of this condition.

It is my understanding that most authors who write GPL'd software do not care who writes the software. I do have GPL'd software, I give it to the public because I think it has "general" use. I don't claim any authorship on any piece of GPL'd code I write, though legally I continue holding the copyright. Imagine what happens when you contribute your code to the GNU project? What is left of your "authority"? It is true, that with GPL, you make a commitment to the commons in a certain way whether you like it or not. (So even Linus or RMS who have egos too big to fit in their brains have to face this commitment)

Non-researchers may have difficulty in understanding why "your name" matters so much in research, but in papers you write your own name on the title of the paper, not "The Internet" as we do in KDE. (To prevent misunderstanding: I contribute to KDE, and I am thrilled to read those fantastic press releases!)

The other points

  1. I don't care if GPL'ing my code allows other researchers do research with my code. It allows them to keep their work secret -> it's short of what's required. That might not be very important for "ls" command but for a work that you're writing papers on it matters more than anything else.
  2. When I say it's the results that matter, I talk of what is published in a paper. So when I say "results", I mean that the license should not prevent me from publishing a paper, present or future. Or more precisely it should not allow certain third parties from "obfuscating" research by doing irreproducible experiments with my software!! The "irreproducible experiment" will be an important term in the forthcoming license ;)
  3. It's of course misreading my definition to claim that "research software is generally useful software". It isn't in any way. It will have 5 users at most in the whole world... And if somebody makes a practical application it will be still open, remember, then after all the research has finished I may want to make a GPL'd release.[*] The name will have been etched in stone by then.

Thanks,

[*] In my case, the subject is parallel computing. The people who can apply the code will need to own expensive supercomputers. I definitely wouldn't want to miss such a business opportunity!! I would rather force them to release "every single modification", OR get a commercial license from me.

You shall talk to Richard Stallman, posted 31 Jan 2002 at 21:17 UTC by atai » (Journeyer)

You should talk to RMS or the Free Software Foundation. They are the experts on this and they may have an answer for you. You do want to use the GPL or a GPL-compatible license so your research can benefit the most people. Again, ask RMS how you can achieve this.

One point, posted 31 Jan 2002 at 21:21 UTC by exa » (Master)

GPL clearly allows "private modifications" and still keep the source code to yourself, while you could say something like "It runs in 1203 seconds using X data". That is the main concern.

GPL, posted 31 Jan 2002 at 22:30 UTC by grant » (Journeyer)

stronger than the GPL?!?

as far as I know it represents the end of the spectrum in terms of freedom - definitely beyond open licenses, it protects the work, its derivatives, and authors from bogus claims which can't be substantiated

how could one possibly hope to prevent private modifications? it is not possible to own ideas unless you choose not to share them. that's the extent to which the term intellectual property can be taken - you can only own what you choose not to communicate.

I understand the scientific publishing community very well - and the GPL is meant to support, extend, and lend organization to the process. Specifically, attribution is required through maintaining copyright notices - there's no need to repeat your own, or your organization's, name so many times in the license, but it seems to occur often, nevertheless.

BSD/artistic/MIT/X and other open licenses allow use (modification) with proprietary works. how would you like your work or its derivative to be patented and/or monopolized? you wouldn't really have a legal recourse beyond copyright unless you use a license more like the GPL. patent law was actually intended to provide such protection, but created unnecessary barriers to innovation through propagation of the intellectual property falsehood. whether you choose the GPL in the best interest of the community or in the best interest of the project makes no difference : innovation and historical preservation are better served through the GPL than patent law.

because of your commitment to the community when using the GPL on your own works, the community tends to disprove false claims even more harshly than otherwise. offer your work to the community, and it will support you in return (by more strongly supporting your work, because the community's stake is greater).

according to the GPL, source is the term used for the state of a work most conducive to modification : for publishing text, then, source and code are one and the same.

the GPL was intended to be general enough to cover many media types, but certainly focuses on software... you may want to check out the Open Publication License used by the United Nations or the GNU Free Documentation License

Yes, stronger than the GPL!, posted 1 Feb 2002 at 00:08 UTC by exa » (Master)

Because scientific practice is a more serious work than programming!

GPL wasn't really designed for "scientific" work.

Would I like an overall more open system where researchers would actively work together to do science? Yes. But currently that's not the case. And it's surely not going to work out like free software does, because of a number of social reasons: It takes an Einstein to come up with a "General Theory of Relativity" and no number of people cannot substitute that. This is of course not the only reason....

Here is how you can write a general clause that covers everything I say:

If one shows the output of a modified version of the software to persons, he should make the entire software, in comprehensible form, freely available to those persons.

The general case I had in mind goes like: If results from a modified experiment setup are conveyed to an individual, that individual shall have rights to access the entire experiment setup, in comprehensible form...... The persons obtaining the experiment setup should be able to reproduce the experiment fully.... No useful information can be omitted from the experiment case....

The general case is harder to come up with but as you can see that is how the license works. "Distribution" is irrelevant. "Giving" the "program" to somebody else is irrelevant. If somebody else accesses the "output" of the program, he can get it in human comprehensible form. That's how you can design a license stronger than GPL!!

As you can see it's possible.

For the special "software" case it also makes a lot of theoretical sense. We have a very good theory of what "input" and "output" are. We define the rights to the source code in terms of "use" with a simple condition that defines use as a causal relation! It is also philosophically much more sound than RMS's license ;)

Limits of copyright-based licenses, posted 1 Feb 2002 at 00:39 UTC by jbuck » (Master)

The kinds of things you want to do are probably not possible with a copyright-based license. Someone who has legally obtained a piece of software has certain fair-use rights, and your copyright doesn't give you unlimited control over them. Copyright gives you a monopoly over modifications to the program or distributing original or modified copies, but it doesn't let you unilaterally impose arbitrary rules. You would either need some kind of shrinkwrap or click-through license, or actually sign papers.

At one time Bruce Perens and a couple of others were talking about trying to tighten the GPL to cover web-based services by calling such services a "public performance". But that seems pretty iffy; the web server is arguably doing a sequence of private performances, one performance for each user.

I don't think that the public is necessarily served if we demand premature publication of source code in all cases (if, the instant I make the slightest modification to a program and show anyone the output I must also give them the entire new code at that point, it may be just too inconvenient for me to help debug someone else's program).

In any case, you don't have to solve everything with a license. Academics are already bound by academic standards to fairly credit those who contribute to their work. Students who violate this and get caught may be expelled, professors may be disciplined or denied tenure.

I think that free software / open source projects often fail to live up to the principle of academic credit, e.g. the Gimp nowhere gives credit to the individual designers of Photoshop (even though, since the Gimp carefully cloned Photoshop's credits box, the coders must have seen the actual names of the designers whose work they were copying without credit). So in this sense, as a researcher you have a reason to be a bit concerned about the open source culture, which tends to credit use of actual source code but not ideas.

In conclusion, if it so offends you that someone might find some way of using your code without compensation, maybe open source isn't for you.

Egos, LGPL, journal, posted 1 Feb 2002 at 13:13 UTC by chalst » (Master)

exa: Very thought provoking article, shame about the double post.

On the friction between academia and the free software world, I quite agree that it is there, but the fault isn't only on the free software side. Though RMS may have an ego problem, in the unlikely case you can convince him that he is wrong he will change. Linus doesn't deserve what you said: he is personally very modest, and his LKML `arrogance' is really good humour. Both of these figures maintain a standard of `accountability' for their creations that I wish was more common in academe.

I don't think the license issue is important, in particular I think the LGPL strikes an adequate balance between the needs of wide dissemination and required publicity. In fact i just invented a new license, the FOLDR (the Free and Open Lincense for Developers and Researchers) which is derived from the LGPL by deleting the theatrics and the annoying `upwardly compatible' clause.

Much more important in my opinion would be a `space' in which academics and free software types can exchange ideas. I think a peer reviewed internet journal to this end would be a good idea, and I have an idea along these lines. I don't want to say too much now (busy with my coming marriage and conference/agency deadlines), but if this strikes a chord, please send me any comments/thoughts/references you might have.

Journals, posted 1 Feb 2002 at 23:52 UTC by exa » (Master)

Well I think academic journals should be open organizations on the Internet rather than cash cows of publishing houses. So this does strike a chord with me.

Printed journals "slow down" science and prevent dissemination of ideas. It's absurd to wait for months in an age when you can perform a banking transaction in 30 seconds.

What I had in mind would be an online journal that requires that the submitted papers contain enough information for exact reproduction of all experiments. If they are talking about a data or program, it should be available freely.

I would like an online journal to be something that every researcher can access without charge and without limitation, and review others' work.

I imagine an agency, a network, that lets scientists review each others work and rank them, organize the knowledge and resources.

Whether that will ever happen, I don't know.

Thanks,

code and research, posted 2 Feb 2002 at 00:41 UTC by mslicker » (Journeyer)

Isn't code very much secondary to Computer Science research? Why do worry so much about someone using your code? It seems the case (I'm not a Computer Science researcher) that someone would look very bad using your results without giving you proper credit.

GPL, etc., posted 2 Feb 2002 at 14:33 UTC by egnor » (Journeyer)

grant: No, that's wrong. It's true that a hosted service which uses GPL software does count as a derived work (as does almost any modification of the software). But you must only give source code when distributing a derived work. If you simply host the derived work on your own computers and distribute only the output of the derived work, then you do not have to distribute the source code to the derived work in question. Running a Web server is not "real-time distribution" of the Web server code; it's merely "real-time distribution" of the Web server's output (i.e. Web pages).

exa: You say "GPL is bound to a particular technology. It depends on the notions of 'binary' and 'source' for instance. Scientific experiments do not always come in the form of UNIX programs." Wait, you were originally talking about research code. Code always has a "preferred form" (loosely known as "source code"). If you're talking about a license for the distribution of non-code scientific material, then that's a whole different discussion.

You say "It's of course misreading my definition to claim that 'research software is generally useful software'". I never said that! What I said is that lots of free software is not "generally useful". Free software doesn't have to be mass-market software -- and most of it isn't! I don't see how the fact that it's only useful to five people alters the equations of freedom. (There is, for example, lots of free software which only runs on "expensive supercomputers".)

No, we don't need a new "license", least of all some kind of horrible clickwrap thing that limits the way you can use things. exa, you have to face the fact that if you publish your ideas, people can use them, sometimes not in ways you approve of. That's part of what "publication" is all about. I for one am thankful that publishers do not have the kinds of rights you are asking for. (This seems to be changing, however, sadly.)

jbuck: Right on.

Bigger problem with copyright, credits and science., posted 2 Feb 2002 at 16:04 UTC by ber » (Master)

There is no need for a "free research license" for software.

Joe Buck is right in that the problem you want to address is hard to solve using a software license. Additionally the Free Software community would like to see more ways to learn and copy information without complicated restrictions.

The scientific community faces a bigger problems with giving credit in a more commercial setting: Not scientific discovery seems to be the first goal anymore, but the money you can earn from your research. Then there is massive information overload. Both problem can only be attacked if we revise copyright laws to have more categories. Scientific work and results have to be treated differently than content for entertainment. The painful grip of copyright on the brain of humanity has to be loosened.

The license will not be more limiting than GPL!, posted 2 Feb 2002 at 20:13 UTC by exa » (Master)

egnor: The license does not have to be more limiting than GPL.

Indeed, it will have a simpler set of rules, remarkable if that counts as a reduction of ad hoc private law which GPL is. Note that GPL uses the copyright law to enforce particular rules that were not present before. Such does my license.

However, as I hinted in previous articles such a license can be crafted maintaining:

  • No Discrimination Against Persons or Groups
  • No Discrimination Against Fields of Endeavor

which are conditions in DFSG if you will remember. I think those are necessary conditions, and in my interpretation GPL cannot uphold them!

The two above rules are those of "equality", and any open system must have rules to preserve equality. Take the case of the web service programmers: they can modify all they want and they don't have to give away a line of code. On the other hand a student who happens to use a free software library for a small application will have to provide his source code. Why should the student have such a responsibility while the former commercial enterprise does not?

In my opinion, it is difficult to uphold equality unless you have objective terms that are not dependent on a particular technology. GPL fails to address a newer technology that is a simple client/server architecture, that's what I meant when I said all research isn't UNIX programs. In the future, there will be radically new forms of computing and GPL will not be able to embrace these new realms.

Therefore whatever legal enforcement is present in the license, it should depend on objective and theoretical criterion with a clear interpretation. The criterion I offer is "output", and the philosophical device for interpretation is "causality". My license could say: anybody who has access to raw or processed output of the machine has rights to the comprehensible form of the computing system. That is I'd replace distribution with output.

To others who said there is no need for an "open research license". GPL does not define what "freedom" and "openness" is. From the viewpoint of the researcher a license that would ensure proper credits and strictly open code would be more appropriate.

I think if there were such a license, more researchers might be willing to open their experiments to others.

Regards,

Credits! Not compensation!, posted 2 Feb 2002 at 20:28 UTC by exa » (Master)

jbuck typed: "if it so offends you that someone might find some way of using your code without compensation, maybe open source isn't for you. "

Assume that I've written a wholly new AI algorithm that achieves a feat previously not thought to be possible. I post a code that does it on the USENET under GPL license. Then somebody takes it and writes a GNOME GUI for using the code in a particular domain. User presses "About" to see the author of the program and sees "Franzo Cecihuli" as the author.

As the researcher, I'd never ever want to confront a situation as the above. 99.9 % of the work is done by me, and Franzo doesn't feel like giving credits? I wouldn't want to trust the good will in others for proper credits, I want to ensure it if it's possible. It should be clearly indicated whom the principal researcher is, and what other codes have been contributed by others. In that case it should read like "Principal Researcher: Blaa Bla.. GNOME GUI by .... ".

Or some wise cracks from "Discovery Solutions Inc." on the sunny hills of california takes the code attaches it to some commercial database and use it to automate a task, and then open up a web service powered by a 5000 node cluster. They don't give the code, or don't have to mention the researcher's name at all!!!

That would be even more unbearable! As the researcher I would like to see my name there _and_ what they have done with the code. I also want to make sure others can see it, too. If they've done any good then my name will lead out of obscurity, and the world will benefit from applications everywhere!!!

I'm all for GPL'ing common codes, I don't give a damn about the credits. Who cares who wrote it if the author does not!!?? Sometimes a BSD like license may be preferable or even putting it in the public domain. We used to have a lot of public domain software before the domination of GPL....

More in a follow-up post....

The most important distinction between research and industry/free software, posted 2 Feb 2002 at 20:33 UTC by exa » (Master)

My advisor told this once to me, after praising me on being a good programmer:

"In the industry or the free software community, it is sufficient to write a code that is overall well. It is perfect to write a code that is better than the average. But in research, you have to do one thing better than everybody else in the world."

That I think would explain why credits are so much important in research.

Honesty is not enforceable, posted 2 Feb 2002 at 21:50 UTC by mslicker » (Journeyer)

This is fact you have to deal with. Especially when society today rewards rampant greed and disonesty. No licence will fix this.

By the way, your research sounds glorified software developement to me. It seems you want to artificially put yourself a level above every one else who does software development.

The Pythagoreans, posted 3 Feb 2002 at 00:31 UTC by nymia » (Master)

Why not try what the Pythagoreans did with their knowledge?

Perhaps you could setup some form of institution or a community that enforces such ownership of knowledge.

Copyright cannot give you what you want, posted 3 Feb 2002 at 00:36 UTC by atai » (Journeyer)

Based on what you wrote, it seems what you want cannot be achieved though a copyright license.

Good research is, by definition, repeatable and verifiable by independent third-parties. So what prevents someone from reading your paper and then clean-room implementing your research results, and useing the clean-room code in whatever fashion without acknowledging you? You have no claim over their code, no matter how you license your code.

What you want can be achieved though patents. But you want to use copyrights, not patents.

So the weakness you attributed to the GPL is not valid. GPL cannot achieve something which the law does not empower it to achieve.

'round and 'round again, posted 3 Feb 2002 at 00:39 UTC by grant » (Journeyer)

egnor : you're talking about use as opposed to actually creating a derivative work. it's a fine line, but there is no difference between linking a local library and a distributed one in terms of copyright.

lower-level services could eventually be available in real-time with some reliability if licenses covered such integration well enough - but it hasn't yet been taken to that point by enough people's minds to be recognized as valid, I suppose

exa : clearly you don't understand the GPL, for it is meant to provide what you continue to describe : (1) attribution and (2) continued openness

grant: you have to read the GPL, posted 3 Feb 2002 at 15:04 UTC by exa » (Master)

And the ways it allows "smart" people to exploit GPL'd code. What is GPL's notion of "derivative work" or "use" of program I ask.

Derivative work is "a work containing the Program or a portion of it".

And GPL says right in the beginning (Section 0) "The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does."

In Section 3 it says that I can copy and distribute the Program "in object code or executable form" provided that I meet certain "copying and distribution" conditions.

You see grant, I can change a "web server", even change it in such a way that its output is completely different, I can run it on my server, and I don't have to give a *single* *line* to anybody else since the output does not contain derivative work, ie no portion of the program itself. I have not "copied or distributed" the "Program" in object code or executable form. The program runs here, the results are sent there. Only the output is sent over the network, not the program or a portion of it. According to GPL I have no obligations whatsoever to offer the source code to anyone.

The situation above is one of those that I wish to avoid decisively.

Without a special license, there is no way that I would distribute research work under GPL. Most of the software is moving to distributing services gradually.

I think I've made clear why GPL is very weak with respect to "technological change". It keeps talking about binary, source code, object code, executable. Those terms have a restricted domain...

A better license should by default state that "any derivative work must be made available to all those who request it under the terms of this license".

atai: no I don't want to achieve that, posted 3 Feb 2002 at 15:12 UTC by exa » (Master)

What you say goes into the domain of patents, and I have no intention to go into that.

If somebody realizes the ideas in the paper himself, fine. There is no problem. What I want is to make sure that all derivative experiments will be made available, rather than preventing others from making experiments!! You see, I don't want some "benchmark paper" guy taking my code, breaking it in such a way that it performs poorly, and then write a paper without releasing the code... That would be disastrous. If I have the license I want, I can just give the code to anybody who wants to study it. And then if I suspect some result I can always try it myself, or somebody else might do that. There would be fewer opportunities for error.

So in reality, this license would increase the number of experiments being made and the overall quality of those experiments, which are indicative of its openness. It doesn't prevent others from replicating the experiments themselves, and it forces all experiments derived from my work to be "open" as well. Fan out.

Restricted rights to make derived works, posted 3 Feb 2002 at 19:51 UTC by chalst » (Master)

exa: I can't say I understand what bothers you about people making a mess of your ideas, but you might take a look at how Bitkeeper keeps derived works on the stright and narrow: derivative works must pass a regression suite to be allowed to be publicly distributed.

Who's choice is it?, posted 4 Feb 2002 at 01:51 UTC by cullenfluffyjennings » (Journeyer)

As a completely tangential issue, you might want to check weather you *can* put out your research code under GPL or anything else. You mention it was done as part of MSc. Most universities this day have some sort of group that does industrial licensing of research. You might be surprised to find that they think they own all your IPR including code.

linking and copyright, posted 5 Feb 2002 at 19:14 UTC by julesh » (Master)

grant said: "it's a fine line, but there is no difference between linking a local library and a distributed one in terms of copyright."

IANAL, but I would say that there are a few distinctions, if I understand what you mean by 'linking a distributed [library]' correctly (I believe you mean making use of functions provided by a remote application in a manner that is in some way analoguos to a remote procedure call or similar mechanism).

Firstly, linking statically to a local library involves making a copy of the code of the function within the library you are using.

If you link dynamically to a library, you will (usually) have to distribute that library with your application, so this also (usually) involves making a copy of the code in the library.

Because of this fact, the above is an action restricted by copyright law for which a specific grant of license would be required. Note that I can do anything I want with dynamic linking on my system as long as I don't distribute the libraries. This, fortunately, is a limitation of copyright law, and can be seen to protect the possessor of a library's freedom to use that library.

Now, with the RPC or whatever you use for the distributed library, no copy is made of the code. Therefore, like in the second dynamic linking case, copyright law does not restrict this action. Anyone can use this for whatever purposes they wish, and no copyright-based license agreement can prevent it.

I think the only way that you could protect against this kind of use would be to declare the output of the work to be a derivative work in itself of the original work. In most cases, this would be rather difficult, if not impossible, to enforce legally. A work must be substantially derived from another to be considered a derivative. The mere fact that it includes a little bit of formatting information or a few constant strings from the original work is not enough.

Note that there is one possible loophole by which all of this could be forced under the stranglehold of copyright law: In order to execute a program, one must make a copy of it in the computers memory. To the best of my knowledge, it has never been tested in a court of law as to whether copyright law would allow the author of a work to make a restriction of use based on this copying: instinct obviously states that it should not be possible, however many proprietary programs include terms in their EULAs that rely on this kind of logic for enforceability, and there are at least some credible arguments to state that the law may be interpreted that way.

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!

X
Share this page