Debian's patent policy FAQ
This document presents information about patents and patent liability useful for developers working on community distributions of Free and Open Source Software (FOSS). By community distributions, we mean collections of free software packages maintained and distributed by organizations composed of volunteers, where neither the organization nor the volunteers seek to make a profit from the activity."
Posted Jul 11, 2011 16:18 UTC (Mon)
by cortana (subscriber, #24596)
[Link] (12 responses)
Posted Jul 11, 2011 16:33 UTC (Mon)
by corbet (editor, #1)
[Link] (4 responses)
Posted Jul 11, 2011 16:48 UTC (Mon)
by tdwebste (guest, #18154)
[Link]
Posted Jul 11, 2011 18:41 UTC (Mon)
by Hausvib6 (guest, #70606)
[Link] (2 responses)
I really want to see the fight between the gorillas and the bears over patents.
Posted Jul 11, 2011 20:43 UTC (Mon)
by nix (subscriber, #2304)
[Link] (1 responses)
Posted Jul 13, 2011 2:54 UTC (Wed)
by Hausvib6 (guest, #70606)
[Link]
So we'll have a great future where being innovative independent little start-up is a curse.
Posted Jul 11, 2011 18:44 UTC (Mon)
by cesarb (subscriber, #6266)
[Link]
I wonder how literate programming combined with an interpreted language fits in this.
Posted Jul 11, 2011 19:05 UTC (Mon)
by rgmoore (✭ supporter ✭, #75)
[Link] (5 responses)
I think they give a reasonable explanation. The idea is that the source code doesn't do anything itself; it's just instructions on how to do something. The end user has to compile it into object code to produce a working implementation of the patent. A description of how to carry out a patented procedure shouldn't be a violation of the patent, or distributing the patent itself would also be illegal. So according to this theory, the worst somebody who distributes source code can be guilty of is contributory infringement, i.e. helping somebody else break the patent. But contributory infringement requires is much harder to prove than direct infringement since it requires knowledge of the patent (or willful ignorance).
I'm not sure how well that theory would work out in practice. IANAL, but I'd think a good lawyer could attack it effectively if he could show that a source-based distribution is more like an infringement kit than a set of infringement instructions. Source code might be like a set of instructions as long as the end-user has to take a set of affirmative steps to build it into object code. But claiming that a source-based distro like Gentoo, where you can install a program from source with a single command, is like a set of instructions is like claiming that a physical product doesn't infringe because it's completely inert until the end user plugs it in.
Posted Jul 11, 2011 19:37 UTC (Mon)
by nybble41 (subscriber, #55106)
[Link] (4 responses)
Object code isn't any more "a working implementation of the patent" than the source code, or the patent itself; the only difference between these forms is the formality of the language. All three forms are mere descriptions of a process, not the process itself or a machine to implement it. You don't have a working implementation until you combine the source or object code with computer hardware and a specific set of inputs. Even then, if the program is running on general-purpose hardware, there really isn't anything novel about it. The software is pure math, and thus not patentable, and the rest (the computer) is only being used for exactly the purpose it was designed for: carrying out mathematical computations quickly and accurately.
> ... like claiming that a physical product doesn't infringe because it's completely inert until the end user plugs it in.
And that would be completely reasonable if the patent only covered the process, and not the machine itself. If you have a patent on the machine it doesn't matter whether or not it's running, as it's the machine's design which infringes, but if you only patented the process then it shouldn't infringe unless it's actually running. In the case of software there is no machine to patent, only the process, which isn't implemented until someone actually runs the software.
Posted Jul 12, 2011 10:36 UTC (Tue)
by Karellen (subscriber, #67644)
[Link] (1 responses)
Interesting.
Would going through the process in your head count as "performing the process"/"running the software"?
Does fully understanding the process necessarily require you mentally perform it?
Does the very concept of a patent, that the details are open *in order that another human can understand it*, become a contradiction-in-terms for process patents, if having another human being understand a process necessarily requires them to perform the process in their head, automatically violating the patent?
(Therefore, I have just proved by contradiction that process patents do not exist. QED :-)
Posted Jul 12, 2011 15:01 UTC (Tue)
by nybble41 (subscriber, #55106)
[Link]
Sure. Which is one reason why purely abstract processes of the sort you can do in your head (e.g. pure math, including all software) should never qualify for patents.
There are other kinds of processes (physical production processes--refining, manufacturing, etc.) which cannot be done in your head. These are more in line with what patents were designed to cover. (Not that I'm saying that such patents are good, only that there is a distinction between physical processes and abstract algorithms.)
Posted Jul 12, 2011 10:37 UTC (Tue)
by pjm (guest, #2080)
[Link] (1 responses)
nybbel41's arguments may be of use to a defense lawyer (so long as the lawyer is also aware of likely counter-arguments, possibly including some of the below); whereas for anyone concerned about infringement, the important thing to consider is that the advice in the Debian patent policy FAQ was prepared by lawyers, whereas nybble41 hasn't claimed any legal credentials or cited any cases or other legal opinions.
Re source vs. object code: the FAQ itself gives reasons for the (tentative) conclusion, which nybble41 doesn't address. It's also a bit harder to argue that x86 object code is provided for explaining the invention rather than execution (practice of the claims).
If the "general-purpose computer" argument is valid, then there's still the problem that Free Software is increasingly being used on non-general purpose machines. One should also consider some of the patents that US courts have upheld, which might make one doubt the safety of even the "general-purpose computer" argument.
Re "isn't implemented until someone actually runs the software", there's in any case still the concern of inducing infringement or contributory infringement.
Again, IANAL; sorry for commenting on things I know little about.
Posted Jul 12, 2011 15:26 UTC (Tue)
by nybble41 (subscriber, #55106)
[Link]
I never claimed to be making a legal argument. If you want to know how a court is likely to rule on the matter, talk to a lawyer--it takes a specialist to deal with the formalized insanity which takes the place of reason in legal contexts. In a courtroom it hardly matters what the reality of the matter is; it's the distorted image in the judge's mind which decides the case.
I was speaking not as a lawyer but as a person who has formally studied the science of computation, and in that context it is obvious that the only real difference between source code and object code is that the meaning of the former is slightly more accessible to non-specialists. Putting aside annotations such as comments and identifiers, which do not affect the operation of the program or (rationally, at least) whether or not it infringes on any patents, the source code and object code are functionally interchangeable.
> ... there's still the problem that Free Software is increasingly being used on non-general purpose machines.
If you're referring to embedded systems, those are still general-purpose computers; don't be confused by a mere change in form-factor. To get away from the "general-purpose computer" argument you would have to actually implement the algorithm directly in hardware, not just run it on an embedded CPU.
Posted Jul 11, 2011 18:24 UTC (Mon)
by gmaxwell (guest, #30048)
[Link] (1 responses)
They have a weird definition of "rarely". While it may be rare relative to all patents filed, in my experience, practically every /interesting/ software impacting patent has office extensions. They're available on request any time office action (like a rejection which is later undone) delays the granting of the patent.
I'm not just nitpicking here: You're liable for infringement that happens prior to expiration, so getting the date a few months off could create a whole world of hurt for you.
Posted Jul 28, 2011 19:18 UTC (Thu)
by SecretEuroPatentAgentMan (guest, #66656)
[Link]
As for the Patent Term Adjustment (PTA) you could check the recent statistics at Patently-O:
This is probably not the time nor place to comment on the document sentence by sentence, still I hope a version 2.0 can be made, quickly.
Source code doesn't infringe
We'd just see more of what we're already seeing - trolls will go after the users, not the developers. Microsoft isn't (as far as I know) shaking down Google, it's shaking down handset vendors who put Android on their devices.
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Source code doesn't infringe
Debian's patent policy FAQ
Debian's patent policy FAQ
http://www.patentlyo.com/patent/2011/07/pta.html
With roughly 80 percent of all patents being extended I would expect that also to be true for software related patents.