[IPOL discuss] Patents and IPOL code

Pierre Moulon pmoulon at gmail.com
Fri Apr 8 13:57:13 CEST 2011


Hi,

What occur for partially patented algorithm ?

I think to all the method that exist and that will be added to IPOL that use
SIFT as a black box for matching.

=> It falls in the Choice 3, with a part of the source code that is
patented.
Or for other choice => Potentially only a subpart of the algorithm could be
published...

Regards,
Pierre

2011/4/8 Jean-Michel Morel <morel at cmla.ens-cachan.fr>

> Dear all,
>
> The final decision lies with the IPOL scientific committee and the IPOL
> editorial board.  We will have to submit them the following alternative
> which summarizes our debate:
>
> Choice 1:  For patented algorithms, the source code will not be available
> to the public, but only to the referees. There will be no online demo for
> the patented algorithms.
>
> Choice 2:  For patented algorithms, the source code will not be available
> to the public but only to the referees. The online demo will be maintained.
>
> Choice 3:  For patented algorithms, the source code will be available, but
> the users will be warned of infringement risks.  The journal will argue that
> the source code is made available as a scientific proof of completeness and
> consistency of the described algorithm. The users will also be warned that
> the source code may be withdrawn from IPOL in case of a request by the
> patent owners.
>
> I can prepare a text to submit this decision to both committees, but I
> would like to have the final draft from José Luis, Nicolas and Pascal for
> the warnings in the source codes (which will be at least delivered to the
> referees).
>
> Best,
> JM
>
>
>
>
> José Luis Lisani a écrit :
>
>> Dear all,
>>
>> I don't think authors should have the right to decide not to make public
>> the code of an algorithm.
>> If it is IPOL's policy to make publicly available the source codes, they
>> have to accept this policy
>> or withdraw the submission.
>>
>> This been said, and concerning the particular submission in which I'm
>> involved, there is a misunderstanding
>> in what has been my intention with respect to the publication of the
>> source code.
>>
>> We are in the process of discussing the convenience of publishing patented
>> algorithms and I have expressed
>> my opinion (I think it is safer for IPOL not to make publicly available
>> the source code), and I have acted in
>> consequence with the publication I'm currently editing (I'm talking in
>> quality of editor, not as a co-author).
>>
>> Since the discussion seemed open to me, I opted for the conservative
>> approach of not providing the source
>> code in the web page. My intention was, once the discussion was over, to
>> follow the policy of IPOL on this
>> matter.
>>
>> After Jean-Michel's contribution to the discuss list it seems that indeed
>> the discussion is closed and the
>> conclusion is that source code should be published and some legal warnings
>> must be added to the code.
>> Therefore, as editor of the paper, I will follow this guideline and I will
>> add a link to the source code in the
>> web site, just as in any regular IPOL paper.
>>
>>
>> Best,
>> José Luis
>>
>>
>> El 07/04/2011 17:09, Jean-Michel Morel escribió:
>>
>>> Dear all,
>>>
>>> Reading José Luis message and his submission together with Juan Biel
>>> Gomila of a patented algorithm from which he has withdrawn public access to
>>> the code I realize that the authors can decide what they want on this
>>> matter. José Luis and Juan Biel Gomila can of course decide not to disclose
>>> the source code at IPOL.  But I certainly do not support this decision, and
>>> I hope it will not happen too often.
>>>
>>> My opinion matters in that I and probably also my institution might be
>>> held personally responsible in case of patent infringement at IPOL.
>>>
>>> I had mentioned that IPOL could give up the disclosure of source code of
>>> patented algorithms. But this was not proposed as a a rule, it was just
>>>  mentioned as a possible retreating line in case of trouble. For the time
>>> being there is no such trouble.
>>>
>>> The  patented algorithms published in IPOL have been disclosed by their
>>> authors themselves in the patent AND in journal papers. The description of
>>> the algorithm that IPOL does is a public discussion of the  public paper and
>>> is therefore not a clear infringement, even if it certainly provides help
>>> for infringement. But any patent discussion does, and there are even
>>> journals dedicated to the analysis of patents, which to the best of my
>>> knowledge are not illegal. As Pascal Getreuer pointed out, the source code
>>> by itself is not a direct infringement, but may be interpreted as a help to
>>> infringement. Thus, it is all a question of degree and of jurisprudence.
>>> From my scientific point of view the scientific paper published by the
>>>  patent authors themselves coul also be interpreted as a help to infringe,
>>> and IPOL simply follows suit.
>>>
>>> IPOL is a purely scientific endeavor. We view the source code as a more
>>> detailed description of a published algorithm, permitting to check that the
>>> algorithm description given by IPOL is complete.  Together with the demo, it
>>> permits users to verify this completeness and the  reality of the
>>> experiments shown by the authors of the algorithm.
>>>
>>> This is why IPOL gives little or no value in the evaluation to well
>>> written code or to a smart implementation: we furnish to the public an
>>> implementation That helps them checking that the algorithm works, and seeing
>>> how it works. This is therefore not aimed at fostering any patent
>>> infringement, but just at informing better the scientific public on the
>>> meaning of an invented algorithm.
>>>
>>> The source code plays therefore more or less the same role as the
>>> publication of a proof for a theorem. It gives evidence linked to the
>>> algorithm, which makes people understand it better. Thus, it is a pity to
>>> withdraw the source code of a patented algorithm, and it will be a pity if
>>> we ever have to stop an online demo.
>>>
>>> On the other hand, I definitely agree with all warning texts that have
>>> been proposed, including the last one written by José. We must indeed inform
>>> the public that certain algorithms are patented, as soon as we get aware
>>> that they are, and we must inform the public that what we publish is
>>> published to foster image science, and nothing else.
>>>
>>> I hope  that José Luis, Nicolas and Pascal will propose a final version
>>> warning text that we could include in the  Zhang-Wu, NL-means, local color
>>> correction, SIFT and ASIFT demos.
>>>
>>> A last point. Since the source code is just on IPOL as an illustration of
>>> the algorithm, we cannot put any license when the source code implements a
>>> patented algorithm. I do not see how we can license a source code, even
>>> conditionally, even with warnings, if we are not the patent owners.  I
>>> consider it safer to just warn users about the patent, as is done in the
>>> text thay you are maturing.
>>>
>>> As Nicolas pointed out we are no patent lawyiers and it is not our job to
>>> enforce patents or to make an inquiry about how far one can go or not go in
>>> the scientific exploration of a patent.
>>>
>>> We are on scientific terrain so far, and nobody has sued us. The day it
>>> happens we will hire a lawyer and find the best strategy. But I simply
>>> refuse to do this preventively.
>>>
>>> At this point, could Pascal propose a new version of the text? I enclose
>>> the last current proposals:
>>>
>>>  ooooooooooo Pascal Proposal ooooooooooooooooooo
>>>
>>>  Copyright (c) 2011, Your Name Here
>>>  All rights reserved.
>>>
>>>  The source file "dmha.c" implements the algorithm covered by U.S.
>>>  Patent
>>>  5629734.  For this reason, the source files "dmha.c", "dmha.h",
>>>  "dmhacli.c" and its binary program "dmha" may only be used for research
>>>  and educational purposes.  Redistribution or commercial use is not
>>>  allowed without the consent of the patent owner.
>>>
>>>  With the exception of the files mentioned above, redistribution and
>>>  use in
>>>  source and binary forms, with or without modification, are permitted
>>>  provided that the following conditions are met:
>>>  (the rest of usual BSD license follows)
>>>  _____________________________________________________
>>>
>>>  For files implementing the patent:
>>>  "may only be used for research and educational purposes" and
>>>  "redistribution or commercial use is not allowed without the consent
>>>  of the patent owner."
>>>
>>>  For all the other files:
>>>  "With the exception of the files implementing the patent mentioned
>>>  above, ..." (BSD/GPL license)
>>>
>>>
>>> oooooooooooo José Luis text oooooooooooooooooooooooooooooooo
>>>  # LEGAL WARNING:
>>>  The source file "localcolorcorrection.cpp" implements the ALGORITHM
>>>  COVERED BY U.S. PATENT 6,822,762. For this reason, this source file
>>>  and its binary program "localcolorcorrection" may only be used for
>>>  non-profit research and non-profit educational purposes. Redistribution
>>>  or commercial use is not allowed without the consent of the patent
>>> owner.
>>>
>>>  With the exception of the files mentioned above, redistribution and use
>>> in
>>>  source and binary forms, with or without modification, are permitted
>>>  under the terms of the GNU General Public License as published by
>>>
>>>  the Free Software Foundation, either version 3 of the License, or
>>>
>>>  (at your option) any later version. The program is distributed in
>>>  the hope that it will be useful, but WITHOUT ANY WARRANTY; without even
>>>  the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
>>> PURPOSE.
>>>  See the GNU General Public License for more details
>>> http://www.gnu.org/licenses/.
>>>
>>> oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
>>>
>>> José Luis Lisani a écrit :
>>>  Dear all,
>>>
>>>  I think that the proposal of Jean-Michel of not making public the source
>>>  codes
>>>  of patented algorithms is the safest for IPOL, in front of hypothetical
>>>  suits.
>>>  Users can still run the algorithms on-line (are we sure that this does
>>> not
>>>  infringe patent laws?) and they will have the guarantee by IPOL that
>>> what
>>>  they run does indeed correspond to what is described in the web site.
>>>
>>>  Moreover, I think that, as a matter of courtesy, the Editor of IPOL
>>>  should send
>>>  the authors of the patent a letter announcing them that there is an
>>>  online version
>>>  of their algorithm. I don't think we should ask for specific consent of
>>>  the authors,
>>>  since, in most cases, they don't even own the invention, their companies
>>>  do.
>>>
>>>  José Luis
>>>
>>>
>>>
>>>  El 06/04/2011 7:42, Jean-Michel Morel escribió:
>>>
>>>  Dear all,
>>>
>>>
>>>  Should we  ask the advice of a patent attorney as suggests Nicolas?
>>>  It  is probably a good idea.
>>>  However, unfortunately, I know some of these guys. They always give
>>>  the most conservative advice. According to the CNRS patent attorney,
>>>  to publish the NL-means code without the owner's consent IS an
>>>  infringement because we help companies to infringe it, even though we
>>>  mention that it is patented.
>>>
>>>  We need a conservative solution showing our good faith. Here is what I
>>>  propose:
>>>
>>>  -systematically, the patented algorithms published in IPOL will be
>>>  sent to their authors as referees. That way we will get a first hand
>>>  information about whether and how the patent is enforced. We will have
>>>  to follow the authors' advice.
>>>
>>>  In case of touchy patent owners, we will only transmit the code to the
>>>  referees under NDA. IPOL will therefore run online a code certified by
>>>  referees, but which will not be public.
>>>
>>>  In other terms IPOL will only publish (actually following its
>>>  definition) "Algorithms", with as a bonus an implementation certified
>>>  by referees, and running on line. The code of this implementation will
>>>  be certified but not public.  The online facility will be installed in
>>>  such a way (limited size, etc.) as to prevent any real or massive use.
>>>
>>>  Still, with all this, a patent owner could object that we are helping
>>>  counterfactors who could do online reverse engineering with our code.
>>>  But, at this point, I think our display of good faith will be solid,
>>>  and it will all remain compatible with the IPOL definition that IPOL
>>>  publishes algorithms and shows how they work.
>>>
>>>  -Meanwhile, we never met the bad guys and we can continue disclosing
>>>  the code. I agree with Pascal's proposal to add the text (enclosed
>>>  below) to every patented algorithm.
>>>
>>>  I do not think we can put any license on patented code without
>>>  incurring in some risk.  Thus I agree that we should separate the
>>>  patented part and put on it the warning proposed by Pascal.
>>>
>>>
>>>  Nevertheless, it is not enough to write "for research and
>>>  educational purposes": we must write "for non profit research and
>>>  nonprofit educational purposes", because there is lucrative research
>>>  and there is lucrative education.
>>>
>>>  I disagree with Nicolas' proposal to put a BSD or other licenses, and
>>>  invite users to decide by themselves if it is legal or not depending
>>>  where they live. Putting a license at this point would be confusing.
>>>  As a publisher we are supposed to know the law. It is our
>>>  responsibility to avoid any ambiguity about rights.
>>>
>>>  Best, JM
>>>
>>>
>>>
>>>
>>>
>>>  Pascal Getreuer a écrit :
>>>  _____________________________________________________
>>>  Copyright (c) 2011, Your Name Here
>>>  All rights reserved.
>>>
>>>  The source file "dmha.c" implements the algorithm covered by U.S.
>>>  Patent
>>>  5629734.  For this reason, the source files "dmha.c", "dmha.h",
>>>  "dmhacli.c" and its binary program "dmha" may only be used for research
>>>  and educational purposes.  Redistribution or commercial use is not
>>>  allowed without the consent of the patent owner.
>>>
>>>  With the exception of the files mentioned above, redistribution and
>>>  use in
>>>  source and binary forms, with or without modification, are permitted
>>>  provided that the following conditions are met:
>>>  (the rest of usual BSD license follows)
>>>  _____________________________________________________
>>>
>>>  For files implementing the patent:
>>>  "may only be used for research and educational purposes" and
>>>  "redistribution or commercial use is not allowed without the consent
>>>  of the patent owner."
>>>
>>>  For all the other files:
>>>  "With the exception of the files implementing the patent mentioned
>>>  above, ..." (BSD/GPL license)
>>>
>>>  So the BSD/GPL only applies to the files that are not covered by
>>>  patents.
>>>
>>>  Having this separate distinction is useful for example for the
>>>  self-similar demosaicking code, where Hamilton-Adams is used as an
>>>  initialization but actually the majority of the code is for the
>>>  self-similar demosaicking algorithm itself.  The license specifies
>>>  just the Hamilton-Adams part as "research and education use only, no
>>>  redistribution" and everything else as GPL.
>>>
>>>  Best,
>>>  Pascal
>>>  _______________________________________________
>>>  discuss mailing list
>>>  discuss at list.ipol.im
>>>  http://tools.ipol.im/mailman/listinfo/discuss
>>>
>>>  _______________________________________________
>>>  discuss mailing list
>>>  discuss at list.ipol.im
>>>  http://tools.ipol.im/mailman/listinfo/discuss
>>>
>>>
>>>
>>>  ------------------------------------------------------------------------
>>>
>>>  _______________________________________________
>>>  discuss mailing list
>>>  discuss at list.ipol.im
>>>  http://tools.ipol.im/mailman/listinfo/discuss
>>> _______________________________________________
>>> discuss mailing list
>>> discuss at list.ipol.im
>>> http://tools.ipol.im/mailman/listinfo/discuss
>>>
>>
>> _______________________________________________
>> discuss mailing list
>> discuss at list.ipol.im
>> http://tools.ipol.im/mailman/listinfo/discuss
>>
>>  _______________________________________________
> discuss mailing list
> discuss at list.ipol.im
> http://tools.ipol.im/mailman/listinfo/discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tools.ipol.im/mailman/archive/discuss/attachments/20110408/1684f24d/attachment.htm>


More information about the discuss mailing list