|
|
Subscribe / Log in / New account

The disabling of hardware codecs in community distributions

By Jonathan Corbet
October 14, 2022
Software patents affect our systems in many ways, but perhaps most strongly in the area of codecs — code that creates or plays back audio or video that has been compressed using covered algorithms. For this reason, certain formats have simply been unplayable on many Linux distributions — especially those backed by companies that are big enough to be worth suing — without installing add-on software from third-party repositories. One might think that this problem could be worked around by purchasing hardware that implements the patented algorithms, but recent activity in the Fedora and openSUSE communities shows that life is not so simple.

In September, the Fedora project changed how it builds the Mesa graphics library, disabling support for the H.264 and H.265 codecs. These formats are heavily encrusted with software patents and have long been difficult to support on Linux systems, though the existence of OpenH264 has improved the situation for many users. In this case, though, the patented algorithms are not being executed on the CPU running Linux; instead, they are run (and accelerated) on a peripheral processor like the GPU. With the change, Fedora users (only of the upcoming Fedora 37 release for now, though that will likely change) have lost access to the acceleration provided by their hardware.

The problem was promptly brought to the Fedora development list, where a number of users expressed their unhappiness at the change. But there was also a certain amount of surprise that Red Hat would not allow code that enables hardware functionality to be shipped; Chris Adams, for example, asked: "But isn't this just providing for hardware decoding, where (presumably) the hardware vendor arranged for whatever needed licenses?". The "presumably", in this case, turns out to be wrong.

The vendors shipping this functionality, it seems, do not feel any need to arrange for the licensing to allow their customers to actually use their products in that way. Instead, the acceleration features in the hardware are viewed as a partial solution that only implements the patented algorithms when combined with software support on the CPU. And, as David Airlie pointed out: it is the provision of a solution that actually works that is seen as using the patented technology:

Think of it like a jigsaw puzzle, where the person who places the last piece in the puzzle pays the license. But then stop thinking of it like that and just assume it's a lot vaguer and way more legally involved than that.

Or, as Tom "spot" Callaway put it:

Mesa was (is) implementing the software parts of H264 and H265 encoding/decoding as part of the Video Acceleration API (VAAPI). This interface tells your GPU (usually, sometimes it is other hardware) to do the hard work of encoding/decoding, but it isn't a black box.

It is a combination of available hardware and software which does the work, and thus, the implementation as delivered potentially infringes.

So, once again, Linux users can't have good things as the result of software patents. Once Red Hat realized that Fedora was shipping code that might possibly be infringing on the codec patents, that code had to be pulled to avoid accusations of willful infringement. After the situation came to light, openSUSE quickly followed suit with a similar change. Some other distributors may not make the same change, choosing instead to continue to ship the affected codecs; Callaway had an explanation for that as well:

They might be too small to sue (remember, expensive to take to court) and/or too small to get any meaningful damages as a result. I mean, Debian is just rolling in money, right? Or the parent company is based in a legal grey zone. *cough*

Companies with sufficiently deep pockets that are not based in legally sheltered areas do not generally feel that they can take the risk, though.

All is not lost yet; Linux users have other codecs available to them. Users who are willing to take the extra step of installing OpenH264 will still have the ability to play back media in that format without infringement worries. Airlie also suggested that having OpenH264 installed might be sufficient to allow the re-enabling of other H.264 codecs on the same system — but it is not clear that any lawyers have signed off on that idea. Some of the disabled functionality may yet show up in third-parties like RPM Fusion as well. Meanwhile, though, this episode is just another reminder of the threats posed by software patents. We are free to write any software we like — but we may not be free to run it.


to post comments

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 15:11 UTC (Fri) by IanKelling (subscriber, #89418) [Link] (11 responses)

Good article. Related link: https://endsoftwarepatents.org/

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 18:06 UTC (Fri) by bferrell (subscriber, #624) [Link] (10 responses)

I knew a developer in the days on the Apple II (not the IIe) who celebrated the legal case decision that allowed software patents.

Up until then, patents could only be granted to physical inventions. "Patented software allows me to get paid for my work for a LONG time after I write it and only I can get paid for something unique". The mouse hadn't extended copyright ad nauseam and had no place to root in the world of software.

While the intent was the individuals would be compensated better, the effect was that any kind of compensation can be turned into a property to be sold and traded.

Now everyone is trying for "do it once and get paid forever" (or a very long time).

An analogy would be a carpenter that builds a house and anytime in the future that it's sold, he get's a royalty of that down stream sale too (I just HAD to say THAT out loud didn't I? Watch and see if it doesn't get tried... Soon)

We're already seeing Tesla pulling silly nonsense on resales of cars with software options that they pull at will and/or charge subsequent buyer for again and again. Books, movies and music... All the same. And the "aware" citizens, they become "pirates" and scofflaws.

Solutions?

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 23:08 UTC (Fri) by developer122 (guest, #152928) [Link] (2 responses)

Thankfully, with codecs it seems this is one of the *very few* areas where voting in the market is working. We're seeing a fairly rapid transition (everywhere that matters) to AV1 to replace these codecs, expressly because of all the fees.

As someone else pointed out (https://lwn.net/Articles/910919/) there's also work underway to kick out the licencing schemes on multichannel audio.

For the larger problem of software patents? Either come up with a way to poison the well or call your congressman, now that it's long-settled law. It's become a similar issue to ludicrously long copyright terms.

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 3:03 UTC (Sat) by linuxrocks123 (subscriber, #34648) [Link] (1 responses)

In the US, Alice v. CLS Bank has dealt a death blow to software patents. We just have to make sure they stay dead :)

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 12:05 UTC (Thu) by esemwy (guest, #83963) [Link]

Well, at least the dumb ones that go, “Do the same thing that people have done for a long time but use a computer instead.”

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 5:38 UTC (Sat) by scientes (guest, #83068) [Link] (2 responses)

I think Bender from Futurama, who runs on the MOS-6502 like the Apple ][, should celebrate. A world of software patents is a world of the robots, for the robots, and by the robots.

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 9:13 UTC (Sat) by nim-nim (subscriber, #34454) [Link] (1 responses)

Naw, it’s world of the lawyers, for the lawyers, and by the lawyers.

Let’s people decide by themselves which one is worse :).

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 11:39 UTC (Thu) by philipstorry (subscriber, #45926) [Link]

I admire your optimism.

But it's actually a world of robot lawyers, for robot lawyers, by robot lawyers.

In eons to come, alien probes will visit what was once our world. And their first interaction will be a query about which licenses they hold, along with a demand for an audit to ensure compliance.

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 5:57 UTC (Sat) by mattdm (subscriber, #18) [Link] (3 responses)

It's worse: the carpenter builds a house, and then gets royalties from anyone who builds any other house in the future if it uses any techniques like "framing" or "joists" or "doors".

But, analogies are dangerous, and this one gets weird because we get into things that ... maybe are okay with patents as they stand. If you design a new construction element that actually is innovative, 20 years might be just fine. Houses last a long time. But with software, especially in the case of codecs as things are progressing today, that basically locks up the entire useful lifetime of the idea.

I think the most reasonable improvements given that we're unlikely to actually get rid of software patents are:

1. Shorten the lifetime. 7 years, maybe.
2. Accidental re-invention should not be infringement, but rather immediate grounds for dismissal of the patent as a natural extension of existing knowledge

And bonus:

OMG do not allow design patents to apply to software.

Beyond or in addition to these:

* If your invention was created with government funding, it should be available for open source projects and academic use without license fees or other obligations. (Yes, I'm biased on this one.)
* Some way in which patented techniques could be used in further inventions without waiting out the original. The current approach, to put it mildly, does not encourage collaboration and building on each other's ideas -- it encourages extending patents with little tweaks that add to the original's effective life. I have no idea how to accomplish this realistically. Maybe simply shortened protection is the best that can be done.
* Perhaps some areas which have other natural drivers for invention just plain shouldn't be patentable. For example, compression will probably always be relevant even as the specific requirements change. Does anyone really think better audio and video codecs wouldn't have been invented if there weren't patents?

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 6:08 UTC (Sat) by mattdm (subscriber, #18) [Link]

And here I'm just making up something that I'm sure people better-qualified than me probably have covered, but:
any technology that becomes an international standard (ISO, IEEE, ITU, whatever) should

1. after some public period, be legally closed to any further "oh wait we have a patent that covers that!" claims.
2. therefore, have an official, reliable, world-wide "patent is done, this is for the world" date

This kind of goes hand-in-hand with some of the other things I listed -- the "no-accidental-reinvention" part could help make this reasonably possible, and from the other direction, the standard itself could cover open source project licensing and so on.

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 13:02 UTC (Thu) by philipstorry (subscriber, #45926) [Link]

I certainly think that the terms for software patents are too long.

My preference is for no software patents, but if we must have them then a much shorter period is required.

Our legislators seem to have forgotten that 20 years was picked as a compromise. In a world of physical manufacturing it was deemed roughly long enough for someone to acquire financing, arrange a location, set up tooling and begin production whilst still giving a period of monopoly on the sale of the invention. Effectively it gave you a head start over the competition.

By the late 90s software was at a point where it was already moving far faster than those 20 years were meant to deal with. Software doesn't require anywhere near as much investment in manufacturing and distribution logistics - a box with floppies or CDs, some booklets and a manual were not that difficult to produce. Many companies were already giving up on that, and by the mid-2000s a lot of software was simply downloaded.

These days we look back on the 1980s ring binders and generous manuals as a luxury of ancient times. If we're lucky then the software we buy has a PDF manual that consists mostly of screenshots. If we're exceptionally lucky then the screenshots in it are for the version we actually bought.

So we're basically looking at a system that was designed to deal with an initial startup period measured in years, but it's being used for an industry which can measure its startup period for products in days. Sometimes even hours.

It was almost defensible in the days of the GIF patent. But today 20 years is just unfit for purpose. 7 years is a more reasonable compromise, but still a compromise.

The disabling of hardware codecs in community distributions

Posted Oct 22, 2022 0:19 UTC (Sat) by himi (subscriber, #340) [Link]

Late to the party, of course, but the "accidental reinvention invalidates" idea is nice, but impossible to make workable in practise - the standard of evidence that would be required to prove that it really was accidental reinvention would be prohibitive. Particularly given the requirement that patents are published.

In theory that should be captured by the question of novelty, something that the patent examiners should be qualified to judge - the fact that they're /not/ is a separate issue from the legal rules around patents.

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 15:21 UTC (Fri) by drago01 (subscriber, #50715) [Link]

"It is a combination of available hardware and software which does the work, and thus, the implementation as delivered potentially infringes."

Without knowing what exactly is done in software and what the hardware is doing that's a very vague statement.

A package with the "disabled functionality" is under review at RPM Fusion

Posted Oct 14, 2022 15:29 UTC (Fri) by knurd (subscriber, #113424) [Link]

> Some of the disabled functionality may yet show up in third-parties like RPM Fusion as well.

Yup, but in case anyone is interested or what's to help, there is a package under review at RPM Fusion that will make it possible to work around this relative quickly: https://bugzilla.rpmfusion.org/show_bug.cgi?id=6426

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 15:44 UTC (Fri) by mfuzzey (subscriber, #57966) [Link] (11 responses)

>Think of it like a jigsaw puzzle, where the person who places the last piece in the puzzle pays the license.

Ok so can't we have that last piece be the individual user (eg by editing a config file or an install time manual option) ?
That way distributions get to ship non crippled code, regardless of the depth of their pockets or their jurisdication and all the individual users will be too small to be a target.

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 16:03 UTC (Fri) by pizza (subscriber, #46) [Link] (10 responses)

They're only shipping software that is useless without hardware.
Other folks only ship hardware (eg the graphics card) that is useless without software to light it up.

This would seem to only affect folks that ship both together in a "complete system", ie pre-installed Fedora Linux?

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 16:55 UTC (Fri) by rgmoore (✭ supporter ✭, #75) [Link] (8 responses)

As a practical matter, I think everyone would probably be liable under the theory of contributory infringement. To quote:

Whoever offers to sell or sells within the United States or imports into the United States a component of a patented machine, manufacture, combination or composition, or a material or an apparatus for use in practicing a patented process, constituting a material part of the invention, knowing the same to be especially made or especially adapted for use in an infringement of such patent, and not a staple article or commodity of commerce suitable for substantial non-infringing use, shall be liable as a contributory infringer.

IOW, they can get in trouble if:

  1. They provide something useful for infringing a patent
  2. It does not have substantial non-infringing uses
  3. They know it can be used to infringe the patent

It's clear from the article that the item can be used to violate a patent and that the distributions know it. So the only question is whether it has substantial non-infringing uses. There are presumably parts of a driver that are useful for letting the user use the non-infringing functions of the card, and those would be legally in the clear, but any function that is only used to start a patent-encumbered CODEC would be infringing.

Honestly, I'm surprised the hardware manufacturers can get away with this. Including a CODEC implementation in hardware has no obvious uses except to infringe the patents. Even if it requires a software component to function, it's hard to see how that isn't contributory infringement.

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 17:36 UTC (Fri) by smoogen (subscriber, #97) [Link] (4 responses)

I expect the hardware device has a license to implement said code. The issue is going to be that the hardware vendor's license is not transferrable to the buyer. Instead there will be some sort of contract saying that there is an assumption that the buyer of said hardware will get a license via a third party or directly from the consortium.

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 19:26 UTC (Fri) by Wol (subscriber, #4433) [Link] (3 responses)

But if the GPU hardware is just a "massively parallel general purpose processor"? It's the perfect device for running this codec on, but it's quite clearly NOT DESIGNED to run it, it's just designed to run that class of problem - like bitcoin mining, or weather forcasting, etc.

Like I said in the OPUS thread, *IF* we can get someone to argue in court "the hardware doesn't infringe, the software is not patentable material, SCOTUS agrees with us, please rule that running the software on the hardware cannot infringe", then we're home and dry. The snag is, it's going to cost a lot of money to get that to court ...

Cheers,
Wol

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 20:23 UTC (Fri) by excors (subscriber, #95769) [Link]

> But if the GPU hardware is just a "massively parallel general purpose processor"?

It's not, the GPUs have dedicated hardware blocks to implement parts of the specific codecs they support, because that's much more power-efficient than trying to do it with the general-purpose compute blocks (in particular I think the entropy coding stage is inherently hard to parallelise and very poorly suited to shader cores) and can also guarantee real-time performance regardless of any other GPU/CPU load.

But Fedora isn't shipping a "complete solution"

Posted Oct 16, 2022 7:21 UTC (Sun) by WolfWings (subscriber, #56790) [Link] (1 responses)

Can disprove this fairly easily actually.

Load up Furmark or some other GPU stress-test program to push the 3D/shader hardware to 100% usage. CPU usage should be near-zero in fact.

Fire up youtube on another monitor, with hardware accelerated video decoding there's still near-zero CPU usage, video is still smooth as silk.

Video decode/encode is separate silicon which is why some cut-rate GPU cores will not have those features, and also why they generally support more decode options than encode: Encoder silicon is FAR more complex and lags generations behind the decoder silicon overall.

If it was GPGPU/CUDA/whatever acronym? You'd have all encoders/decoders all the time on anything that supported advanced enough pixel shaders.

But Fedora isn't shipping a "complete solution"

Posted Oct 16, 2022 16:24 UTC (Sun) by farnz (subscriber, #17727) [Link]

Note that post-processing and encoders (but not normally decoders) do often use programmable shaders in conjunction with fixed hardware. For example, Intel's current GPUs use programmable shaders for parts of (at least) Blending, Colour Space Conversion, Rotation, Scaling, and Color Fill during post-decode media processing, while the encoders can use programmable shaders to enhance the output for at least H.265 (HEVC), H.264 (AVC) and MPEG-2.

So even if you see some slowdown in your GPU stress-test program when you also do hardware-accelerated video decode, this is not indicative of the whole process being done on the programmable shaders - it's just that bits of the process are done by shaders and other bits by dedicated hardware.

But Fedora isn't shipping a "complete solution"

Posted Oct 15, 2022 14:33 UTC (Sat) by farnz (subscriber, #17727) [Link]

To answer your last paragraph, the patent pools usually sell a restricted licence relatively cheaply for hardware that is non-functional without a software component.

So the hardware has a licence which protects the hardware vendors, but not distributors of software to enable the use of that hardware. You need a separate licence to distribute the software that enables use of that hardware, and if you don't, you're engaging in contributory infringement, as is the user who assembles the parts together unless you have a complete licence.

There's also a licence available for distributing software to resellers that puts the burden of sorting the final licence out on the reseller - so that it's possible for (say) AMD to have cheap licences for their hardware and drivers, and to put the final expensive licence burden on a company selling AMD chips to end users.

But Fedora isn't shipping a "complete solution"

Posted Oct 17, 2022 23:04 UTC (Mon) by linuxrocks123 (subscriber, #34648) [Link]

Fedora isn't typically sold, though, so that wouldn't apply. There's also this decision, which concerns the definition of "component", and which ruled that software code is _NOT_ a component of an invention:

https://en.wikipedia.org/wiki/Microsoft_Corp._v._AT%26T_C...

That's related to "component" as used in 35 USC 271(f), and contributory infringement is 35 USC 271(c), but I think it would be pretty difficult for someone to successfully argue that the same word three paragraphs up should be defined differently.

But Fedora isn't shipping a "complete solution"

Posted Nov 2, 2022 22:37 UTC (Wed) by fest3er (guest, #60379) [Link]

To be effective and accepted in the long term, law must be interpreted as narrowly as possible. To do otherwise would, in time, make every action illegal.

According to the quoted text, the key points are:
(1) offers to sell, OR
(2) sells, OR
(3) imports into the US

Thus, if one does not offer for sale, does not sell, AND does not import into the US said bits and pieces, one will not be liable as a contributory infringer.

By the quoted text, one who independently writes code within the US and gives it away for free cannot be held liable as a contributory infringer.

But Fedora isn't shipping a "complete solution"

Posted Oct 14, 2022 18:55 UTC (Fri) by notriddle (subscriber, #130608) [Link]

Wouldn’t shipping software like that violate Fedora’s Free Software guidelines?

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 16:43 UTC (Fri) by q_q_p_p (guest, #131113) [Link] (1 responses)

Gentoo solves this problem nicely - it leaves the choice to have these codecs or not to the user, like real free software should do ;)

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 20:29 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]

> Gentoo solves this problem nicely - it leaves the choice to have these codecs or not to the user, like real free software should do ;)

Real free software as defined by licensing always allows the users to modify whatever they want and distros removing the codes in their official repository still don't in any way inhibit users from installing the codecs from a different repository, so this isn't the salient difference here.

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 17:41 UTC (Fri) by jhhaller (guest, #56103) [Link]

Hmm, sounds just like the Opus patent pool news article comments. Even if providing the software is not providing something which provides a patented system, it could be seen as contributory infringement. One wonders how far that goes in other areas. For example, does the work to support HDR video output potentially infringe on any patents media players or their vendors have filed for their HDR implementations? The company I work for has negotiated license agreements with automobile manufacturers for integrating mobile modems into vehicles, presumably for patented functionality enabling integrating patented mobile internet capability into the car. This presumably means the wireless functions were not exhausted by building the modem. If one installs a 4G/5G modem into a laptop and streams a MPEG2 video through a web browser, it's conceivable that that combination could be a patented part of 4G/5G standard, even though the MPEG2 patents have expired. The modem maker doesn't want to provide the license for integrating their product into a media player. Would that make the browser supplier liable? The computer manufacturer who provided IO connectivity allowing connecting a modem? While the end user is the one who put everything together, law generally allows "internal use" of patents as non-infringing, so one can't go after them.

I don't know the way out of this morass. If we continue to allow patents on devices, which have part of their functionality provided by software, either people who sell software allowing patented capabilities to be installed by the customer avoid paying the license, or an enforcing license check has to be built into software. The former does not encourage investment in expensive capabilities, the latter is incompatible with FOSS, and particularly incompatible with GPL licensing. It's not just a threat to FOSS, but also to proprietary vendors who have no idea how their end customers will assemble their systems. Even if a license was paid for the video codecs, integrating it with a 4G/5G modem could be contributory patent infringement.

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 18:04 UTC (Fri) by flussence (guest, #85566) [Link] (2 responses)

On the one hand, that'd be pretty annoying if I used those distros.

On the other, I don't need GPU H264 encoding at all - my last system upgrade was instead partly motivated by encountering a Youtube VP9 it couldn't keep up with.

On the other other hand, H265 is very much a "the emperor has no clothes" situation. I've never encountered a H265 file in the wild; no content hosting site will give it the time of day, and the only real user I know of is some guy who possesses multiple hard disks' worth of TV shows and a healthy disregard for the legality of any part of the process. It's a format for patent trolling, it gets beaten by 10bit H264 in benchmarks, and nobody significant uses it (yeah Apple does, but Apple also uses HFS+ and flat geometric design with Helvetica). I'd prefer software to throw an error upon encountering the format out of pure spite. Creating content in it ought to carry a humiliating stigma.

The disabling of hardware codecs in community distributions

Posted Oct 16, 2022 7:34 UTC (Sun) by koenkooi (subscriber, #71861) [Link]

Iphones default to HEIF pictures and HEVC movies, my Canon photo cameras also use HEVC for video as do other brands. Android 10 introduces HEIF+HEVC support and starting with Android 12 various OEMs changed the default capture format to be HEVC.
So there are a lot of consumer generated h265 files in the wild, but mostly inside devices and ecosystems where it Just Works(TM) and the royalties have already been paid.

The outlier is the photo camera situation, working with 10-bit YUV422 HEVC files is painful since there isn't a lot of hardware support for it outside of Apple Silicon, running ffmpeg+vidstab to get rid of camera shake will take hours for a 30 second video clip, on an 8-core Ryzen. This is mostly self inflicted damage, since I don't know what I'm doing when it comes to video and just select "highest quality" everywhere and expect the tools and hardware to catch up :)

The disabling of hardware codecs in community distributions

Posted Oct 16, 2022 16:27 UTC (Sun) by farnz (subscriber, #17727) [Link]

10 bit H.265 beats 10 bit H.264 in quality per bit benchmarks for both 10 bpc and 8 bpc content. 10 bit H.264 beats 8 bit H.265 in quality per bit benchmarks for 8 bpc content, which can be surprising until you look at how both codecs blend reference frames and the residual into the decoded frame.

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 19:07 UTC (Fri) by Foxboron (subscriber, #108330) [Link] (14 responses)

> I mean, Debian is just rolling in money, right?

Well, they have 700k USD on hand according to SPI. All of that is liquid assets.

https://www.spi-inc.org/treasurer/reports/202207/#index32h4

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 19:33 UTC (Fri) by Wol (subscriber, #4433) [Link]

The lawsuit alone would burn that up, most likely ... even with assets like that, Debian is a "man of straw", not worth suing. It sounds a lot, but it really isn't ... just about enough to pay one year's salary for the CEO of a mid-size company.

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 14, 2022 20:21 UTC (Fri) by rahulsundaram (subscriber, #21946) [Link]

> Well, they have 700k USD on hand according to SPI. All of that is liquid assets.

Yes, this actually reinforces the point that Debian risk profile is very different. The optics of going after large commercial organizations with other large commercial customers and a chance of considerably better payout is much more enticing than going after a non profit foundation where you can transparently see the upper limits of it.

The disabling of hardware codecs in community distributions

Posted Oct 16, 2022 11:29 UTC (Sun) by pabs (subscriber, #43278) [Link] (11 responses)

Debian's patent policy (with advice/FAQ provided by the Software Freedom Law Center (SFLC)) is here btw:

https://www.debian.org/legal/patent
https://www.debian.org/reports/patent-faq

The disabling of hardware codecs in community distributions

Posted Oct 16, 2022 20:32 UTC (Sun) by Conan_Kudo (subscriber, #103240) [Link] (10 responses)

That policy does not jive with what Debian is doing right now. Debian explicitly re-enabled the codecs in Mesa, which makes it work for AMD users out of the box now. They ship x264, x265, and ffmpeg with all this stuff enabled.

Just based on reading that policy and looking at Debian actually does, I don't think anyone actually follows the policy.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 1:47 UTC (Mon) by pabs (subscriber, #43278) [Link] (9 responses)

I note that the policy was mentioned in the bug for that commit, but a user was able to convince the maintainer that the previous default for mesa should override the policy. I've passed this article to the maintainer btw.

https://bugs.debian.org/1016687

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 3:43 UTC (Mon) by timrichardson (subscriber, #72836) [Link] (8 responses)

thanks for the link. The debian maintainer was convinced that the previous situation was compliant, not that the previous situation was ok simply because it was the status quo. The argument considered the legality of the driver in debian called intel-media-va-driver, and considers that the AMD situation is the same. That is, the use of the patented logic is not enabled merely by mesa alone, or by the intel driver alone. Red Hat's developer, presumably based on advice from Red Hat legal, came to a very different conclusion from the same starting point, at least for Mesa. He didn't seem very convinced or convincing: it's a jig saw puzzle with one piece missing, but actually it is something else.

It gets confusing pretty fast to me: it seems that Fedora regards the fully enabled intel driver as non-free because of the patents, but includes it in repositories anyway ( tagged non-free to enable to enable user choice), but mesa + H264 which has a risk of infringing those same patents in combination with that non-free driver, will not be shipped. If one is ok, why not the other, what is the difference? The debian maintainer seems to conclude that (1) there is no difference and (2) each of the components on its own is not infringing. To me, this seems more consistent than the Fedora/RedHat position.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 4:56 UTC (Mon) by zdzichu (guest, #17118) [Link] (2 responses)

The Fedora comment is untrue. Package libva-intel-driver lives in RPMFusion repository, which is 3rd party repo, not a part of Fedora. There isn't anything like "non-free" in Fedora.
Also, RPMFusion is in the process of packaging mesa-freeworld, with limitation described in this article reverted.

I've found interesting tidbit in the bug you linked: “If Debian doesn't ship the firmware, there is no issue.”
Given recent decision to ship firmwares in Debian, doesn't it create a legal problem?

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 7:47 UTC (Mon) by Conan_Kudo (subscriber, #103240) [Link]

And actually, Intel makes it possible to build the VAAPI driver without support for encumbered codecs. That version is being worked on to bring into Fedora proper.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 10:43 UTC (Mon) by timrichardson (subscriber, #72836) [Link]

Thank you for the Fedora correction. In that case, Fedora is consistent.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 7:41 UTC (Mon) by Conan_Kudo (subscriber, #103240) [Link] (4 responses)

Debian recently changed to ship firmware on the media and install by default, so that changes the dynamics quite a bit. Putting that aside, Debian provides plenty of software-only codecs that are patent encumbered. According to the linked policy, Debian shouldn't be doing that. But since it does, it's clear nobody follows that policy.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 9:49 UTC (Mon) by amacater (subscriber, #790) [Link] (3 responses)

The Debian firmware GR is about non-free firmware (for things like WiFi) and not "all of what's in Debian non-free" - that's the *whole* point. The installer will install the firmware - see the contents of firmware-linux-nonfree for things that install in /lib/firmware - but will also offer the end user the chance to back out and install only a free system.

This is shipping binaries provided by the vendors and licensed as distributable. That's the primary reason that they're non-free. Debian doesn't have source, can't fix problems if they arise etc. That's the rationale for segmenting Debian's non-free into firmware (which almost all Linux distributions will then distribute identically without significant concerns) and the rest of Debian-provided non-free firmware.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 10:21 UTC (Mon) by Conan_Kudo (subscriber, #103240) [Link]

Yes, but that also includes the AMD and NVIDIA GPU firmware, which are required to make the open drivers for both GPUs functional. That firmware means that Mesa's VAAPI drivers will work too.

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 13:13 UTC (Mon) by amacater (subscriber, #790) [Link] (1 responses)

Drat it - one word too many left in. It should read with something of the sense given below:

That's the rationale for the split of Debian non-free into two parts - Debian non-free firmware as one part - and the *rest* of what is currently in Debian non-free as the other.

In the post of mine that this is a direct reply to, I left the word firmware on the end of the final sentence in error and made it a nonsense.
How come you never notice these things when re-reading the preview?

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 13:25 UTC (Mon) by Conan_Kudo (subscriber, #103240) [Link]

How come you never notice these things when re-reading the preview?

It's because you fill in everything in your head as you re-read and process it. You only notice once you have some distance, and by then it's too late.

How about in Europe?

Posted Oct 14, 2022 21:20 UTC (Fri) by fraetor (subscriber, #161147) [Link] (9 responses)

I've never understood how this kind of software patent issue applies to those of us in Europe, where software patents are not grantable[1]. Can you still be taken to court by a USA based company if you are entirely based in Europe? Or are you liable if you distribute to users in the USA?

[1] Article 52, section 2c <https://www.epo.org/law-practice/legal-texts/html/epc/201...>

How about in Europe?

Posted Oct 14, 2022 22:03 UTC (Fri) by Wol (subscriber, #4433) [Link] (5 responses)

You can be sued in the European Court, with a software patent granted by the European Patent Office.

They're illegal in the States just as much as they are over here. But our Patent Office is just as eager to grant them as the USPTO.

The lawyers have successfully argued that Black is White such that "software patents as such" (or whatever the wording is) are Unicorns in Europe. Just like in the US software patents don't infringe the ban on patenting maths, or patenting the problem and closing down all other possible solutions ...

That's where we have a problem, it's persuading the Courts that software patents are actually in breach of all precedents that require patents to be a SOLUTION to a problem, not just a statement of the problem.

Cheers,
Wol

How about in Europe?

Posted Oct 14, 2022 23:40 UTC (Fri) by wahern (subscriber, #37304) [Link] (3 responses)

It's not evident from the FOSS community's bubble but there are legions of economists, lawyers, MBAs, and politicians that are die-hard absolutists regarding the beneficial efficacy and importance of patents to "innovation". Moreover, to these people all attempts at distinguishing software patents (or any class of patents) are intrinsically non-sensical.

They're not a majority. Most people aren't apt to understand or analyze the patent system in such stark, rigid terms. But in the context of patent debates their ideological fervor helps them punch well above their weight. And they have particularly strong representation among lawyers as well as in many economic policy circles, which is why they're able to stay so well entrenched.

Also, the Japanese business community and government see patents as one of their principal competitive advantages, and I suspect they discretely but firmly communicate their position to EU, US, and other policy makers. This is true for other countries like Israel and Korea, but Japan is far more important economically and politically, especially to the US in particular, and maintaining a patent regime to Japan's liking might very well be a key quid pro quo in US-Japan relations.

How about in Europe?

Posted Oct 15, 2022 4:29 UTC (Sat) by fw (subscriber, #26023) [Link]

In Europe, you also have the viewpoint that the ability to obtain patents is a natural right, related to the right to personal property. It's markedly different from the U.S. view, where the Constitution specifically limits copyright and patent legislation “To promote the Progress of Science and useful Arts”. Under the property rights theory, it does not actually matter if patents foster innovation or have other economic benefit.

How about in Europe?

Posted Oct 21, 2022 16:12 UTC (Fri) by mattdm (subscriber, #18) [Link] (1 responses)

> It's not evident from the FOSS community's bubble but there are legions of economists, lawyers, MBAs, and politicians that are die-hard absolutists regarding the beneficial efficacy and importance of patents to "innovation". Moreover, to these people all attempts at distinguishing software patents (or any class of patents) are intrinsically non-sensical.

Yes, very much this. When my kids were small and I basically lived at the playground near my house, there was a woman also often there who worked for a firm specializing in patents (both filing and litigation, although she was on the filing side). She loved her work, and felt strongly that what she was doing was important, good work that helped inventors and protected innovation. We had some good conversations, and I don't think convinced each other at all -- but it was eye-opening. And I hope likewise, as the angle we* see it from was totally surprising to her.

> They're not a majority. Most people aren't apt to understand or analyze the patent system in such stark, rigid terms. But in the context of patent debates their ideological fervor helps them punch well above their weight. And they have particularly strong representation among lawyers as well as in many economic policy circles, which is why they're able to stay so well entrenched.

This makes sense structurally: the jobs in software patents are, you know, rather dependent on software patents _being a thing_. There's no money on the other side.

----

* I'm going to go ahead and lump all of us here together generally, since... it's a pretty bubbly bubble!

How about in Europe?

Posted Oct 24, 2022 11:35 UTC (Mon) by paulj (subscriber, #341) [Link]

When society is generally so structurally invested in the existence of some system of sectoral governance, such that removing its influence from your sphere of concern (where it may be more malign than benign) is essentially impossible, then you have to look instead at how you can use that system against itself to protect your concerns.

Basically, seems we need to have strong "open source" patent pools. E.g., perhaps ones like http://www.openinventionnetwork.com/ .

How about in Europe?

Posted Oct 15, 2022 20:14 UTC (Sat) by SLi (subscriber, #53131) [Link]

There's some genuinely confused and headache-inducing case law on this. The European Patent Court has explicitly ruled that the "software as such is not patentable" clause is merely a restriction on how the claim must be drafted. That, I believe, is binding patent law all over the EU. So the reality is that, unlike what is the common thought, EU is a software patent wild west, and that is binding in the entire EU. Certainly it is my impression that EU is much more tolerant towards software patents than the US. So I'd wish the myth of no software patents in the EU died.

Moreover, the same ruling took the interesting position that while an invention needs to be "technical", "novel" and "inventive". However, these all need not all exist at the same place in the "invention": In the language of the decision, "the identified further technical effect
need not be new" (EPO Opinion G03/08). You would be able to patent, for example, novel and inventive, ways to present information (no technicality involved) on a computer screen (sufficient, because the computer, while not novel, is technical)—the same would apply to software on any technical storage medium.

At times, it becomes even crazier when lawyers lawyer. In the UK, the highest court arrived at a conclusion that advocates for less patentability for software: That software "is normally" not patentable, unless it has a concrete effect on the world. Sounds good? Well, the example they chose (and ruled based on) was that a memory allocator would be patentable since it "enhances the computer", making it better in performing its function, thus essentially making it a better computer.

I actually secretly suspect some of the UK bureaucrats understood the absurdity of this position when they wrote position papers at the EU level, but they just were mandated to advocate for that position...

How about in Europe?

Posted Oct 15, 2022 4:18 UTC (Sat) by fw (subscriber, #26023) [Link] (2 responses)

The claims usually cover a processor doing something, or storage media containing something. Technically, they are not about software. Even in the U.S., I don't think anyone writes software-only claims these days.

What can be patented is not actually that interesting, it matters more how infringement occurs. And the current situation in Europe is that software vendors can be liable even though infringement only happens once their software runs on some piece of hardware they have not made, sold, or provided in any other way. This is not as preposterous as it may sound because it's not uncommon for infringement to occur only after multiple hardware parts and manufacturing steps have been combined along a supply chain. Without some form of liability for contributory infringement, it would just be too easy to avoid most patents.

How about in Europe?

Posted Oct 15, 2022 15:05 UTC (Sat) by Wol (subscriber, #4433) [Link] (1 responses)

> This is not as preposterous as it may sound because it's not uncommon for infringement to occur only after multiple hardware parts and manufacturing steps have been combined along a supply chain. Without some form of liability for contributory infringement, it would just be too easy to avoid most patents.

Except that, combining a patentable item with a non-patentable item IS NOT PATENTABLE.

So, if software is not patentable, running software cannot infringe.

We just need to persuade the courts this is settled law ...

Cheers,
Wol

How about in Europe?

Posted Oct 17, 2022 20:52 UTC (Mon) by zoobab (guest, #9945) [Link]

Software Patents in Europe are coming back through the installation of the Unified Patent Court, where Germany should ratify and launch the court in mid-December.

This court will replace National Courts, where some of them have hostile jurisprudence against software patenting, like the French Courts.

The problem is that there won't be an appeal possible to the European Court of Justice (CJEU) possible, since the EU is not part of the Unified Patent Court Agreement (UPCA), it's a deal between some Member States and not the EU.

So patent law is left to specialized patent courts, highly dangerous for society as we saw with the US example, before the Supreme Court intervene in the Alice decision.

The Unified Patent Court project is also legally flawed, as it claimed to be a 'common court', while the jurisprudence of the CJEU says 'common courts' need to have a link with National Courts, the guardians of EU law together with the CJEU:

https://ffii.org/unified-patent-court-wont-survive-a-lega...

We should convince a government to challenge this 'common court' and the validity of the Brussels1 regulation at the CJEU, preferably before this treaty enters into force.

Governments are also magouilling the entry into force to ignore the requirement of the UK:

http://patentblog.kluweriplaw.com/2022/04/21/upc-four-rea...

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 5:40 UTC (Sat) by scientes (guest, #83068) [Link]

If my King did not issue the patent I don't consider it law, unless it was personally reviewed by either Kafka or Albert Einstein.

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 9:56 UTC (Sat) by nim-nim (subscriber, #34454) [Link] (29 responses)

Patents are incompatible with an high-tech economy.

High tech means exactly that, you are combining lots of technology in a single product. If each part has several patents attached, you quickly come to the situation where :
– it is not possible to define a royalty cost per patent low enough, that the sum of all them won’t dwarf the potential benefits of manufacturing and selling the result
– even if it was possible, the administrative costs of tracking what bit implements what patent, what patent is still active and what patent has expired, and to whom the micro-royalty should be paid to would dwarf the royalty costs
– that means you stop dealing with individual patents you start dealing with patent pools (obscuring all the expiration mechanisms the legislator insisted on)
– that means you induce the creation of huge cartels, which are the only ones able to handle those pools (and they don’t even pay individual patents, they don’t even bother to track where they use them, they threaten one another with MAD and don’t pay an actual dime to anyone)
– that means you kill all the small actors, the ones most creative to start with
– the end game as we’ve seen for covid vaccines is government buyer associations negotiating prices directly with producer cartels, a top-heavy organisation which would be utterly incapable of innovating long term

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 10:00 UTC (Sat) by nim-nim (subscriber, #34454) [Link]

(that’s one huge reason high tech manufacturing moved to regions like China, where patent enforcement is limited).

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 21:53 UTC (Sat) by khim (subscriber, #9252) [Link] (2 responses)

It's not just contemporary high-tech. The really fascinating (and easily verifyable!) story about patents are chemical industry. You can read about that here.

> In 1862 British firms controlled about 50% of the world market, and French firms another 40%, Swiss and German companies being marginal players. By 1873 German companies had 50% of the market, while French, Swiss and British firms controlled between 13% and 17% each. In 1913 German firms had a market share of more than 80%, the Swiss had about 8%, and firms in the rest of the world had largely disappeared. During this entire period there was no patent protection at all in Switzerland, while in Germany processes become patentable in 1877 but products did not. In France, the U.K. and the U.S. both products and processes had been patentable all along.

The real trouble with patents is just the fact that while long-term effects are quite obviously destructive short-term effects are the opposite: they make freshly-patented endeavours more lucrative and bring more money to the industry. But they also push out the smaller companies to other countries where innovations continue and eventually industry implodes.

But this process takes decade or two… and legislators never look on such timescales, they usually look on effects which happen on 3-4-5 years.

The disabling of hardware codecs in community distributions

Posted Oct 16, 2022 10:05 UTC (Sun) by Wol (subscriber, #4433) [Link] (1 responses)

Another clearly identifiable problem with patents is steam engines ...

James Watt patented the LOW pressure steam engine, and refused to licence the tech to anybody who wanted to build a HIGH pressure engine.

Fast forward till the patents expire, and locomotives appeared almost immediately. Completely impossible with low pressure engines (not enough power-to-weight), as soon as people could build high pressure engines they got so much better power-to-weight ...

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 17, 2022 23:15 UTC (Mon) by linuxrocks123 (subscriber, #34648) [Link]

This example does not appear to be valid: https://econfaculty.gmu.edu/pboettke/workshop/Fall2009/Se...

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 1:03 UTC (Tue) by Paf (subscriber, #91811) [Link] (24 responses)

It’s weird - a lot of people outside of software think they’re essential to one.

I think it’s more situational.

It is absurdly difficult and expensive to *create* a drug (the drug industry R&D budgets are a real thing and they don’t throw that money away for fun), but the cost to *produce* is generally trivial. COVID vaccines are a great example - Moderna spent billions before creating their first viable product.

It’s a situation that absolutely cries out for something like patents - otherwise we’d need an entirely new model for development of drugs. And that could be a thing, but it would require huge changes.

Software patents are a far more difficult issue, obviously. Most of them seem to be bad.
I think we just don’t have terribly good answers to the *why* of patents that cover enough of technological activity, particularly software.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 9:45 UTC (Tue) by nim-nim (subscriber, #34454) [Link] (6 responses)

One reason it’s absurdly difficult to create new drugs is that big pharma players go out of their way to avoid the patents of other players. The patent lock is such a lot of money is expended in inventing new square wheels.

Drug creation is more and more akin to software, with new molecules and processes derived from previous molecules and processes, and a derivation cadence completely incompatible with patent duration.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 18:44 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (5 responses)

> One reason it’s absurdly difficult to create new drugs is that big pharma players go out of their way to avoid the patents of other players. The patent lock is such a lot of money is expended in inventing new square wheels.

Sorry, but this is total nonsense. Drug patents in general are _very_ specific. So they will apply only if you are working on improving an existing drug.

New drugs are expensive because they _mostly_ _fail_. Around 90% of drug candidates fail during clinical tests.

It's that bad. And if anything, it's getting worse.

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 7:54 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (4 responses)

Sorry, but no. The stories of companies pivoting from research on other illnesses to covid, or trying to prove a variant of existing patented drugs works on something else, could have been written about software startups.

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 11:29 UTC (Wed) by Wol (subscriber, #4433) [Link]

Bear in mind, in the UK, a lot of companies got given contracts that effectively gave them a blank cheque to work on CoVid. I don't know all the details, but it sounds like it was along the lines of

Pharma: we've got this idea we think might work ...
UK Gov: Okay, try it, here's a contract for 10 million doses whether it works or not ...

That's one of the reasons our vaccine roll-out was so fast - we ordered huge quantities before other countries really woke up, so we were at the head of the queue once pharma started delivering.

So a lot of money was "wasted" on vaccines that didn't work. But how much money was *saved* because we were well ahead of the curve getting our people vaccinated? Personally I think it was money well spent ...

So yep, the companies were throwing research at it, but it wasn't research THEY were paying for.

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 19:34 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

Which stories?

Repurposing existing drugs does happen, but it's a really minor activity. Simply because there's not too much to repurpose.

You can look at the list of FDA approved drugs: https://www.drugs.com/newdrugs.html

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 20:52 UTC (Wed) by Wol (subscriber, #4433) [Link] (1 responses)

I think you'll find re-purposing is major business ... the classic example of course is that little blue pill ...

But certainly for the illness I'm involved with (Parkinsons), probably the majority of the current research bill is looking for drugs to re-purpose ...

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 19:38 UTC (Thu) by Cyberax (✭ supporter ✭, #52523) [Link]

> I think you'll find re-purposing is major business ... the classic example of course is that little blue pill ...

Viagra is not an example of a repurposed drug. It's an example of a new drug that failed in its clinical trials (against angina pectoris). It just so happened that its side effects were interesting enough in their own right.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 10:16 UTC (Tue) by paulj (subscriber, #341) [Link] (7 responses)

Moderna got government grants to fund development of their mRNA vaccines. They also benefited from research at various universities world-wide on such technology, paid for... by government grants.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 18:45 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (6 responses)

> Moderna got government grants to fund development of their mRNA vaccines

Moderna spent nearly 10 years working on mRNA vaccines on its own, using about $5B of private funding. Government funding came only during the last stages of the clinical testing and manufacturing scaleout.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 19:09 UTC (Tue) by paulj (subscriber, #341) [Link] (5 responses)

I don't think your statement is correct. Moderna received a number of grants from DARPA to work on mRNA. Since at least 2013.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 19:12 UTC (Tue) by paulj (subscriber, #341) [Link] (1 responses)

Oh, and BARDA too. At least hundreds of millions from BARDA - before Covid.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 19:16 UTC (Tue) by paulj (subscriber, #341) [Link]

A PR from Moderna mentioning /one/ of the BARDA grants. https://www.businesswire.com/news/home/20160907005580/en/...

They have also either received grants from other governments, or partnered with academics funded by other governments too.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 19:22 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

The ones that I found were around $25 million. The private funding over the years was over $3B: https://www.crunchbase.com/organization/moderna-therapeut...

Keep in mind, that most of these investments were during the phase when it was not clear at all if mRNA is going to even work.

Sorry, but the story of "government funded mRNA vaccines and is now screwed by greedy pharma companies" is not even close to the truth.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 20:27 UTC (Tue) by Wol (subscriber, #4433) [Link] (1 responses)

And wasn't a lot of the early work done by a lone female researcher, who missed out on promotions and nearly bankrupted herself, because she believed in the technology when a lot of people *wouldn't* back her?

I may be muddling up my stories, but I think it's this one, it IS the classic case of a lone researcher burning themselves out both emotionally and financially, and the vultures pile in once they realise it does actually have potential ...

She's *now* financially well off, but only because various people thought she deserved it and rewarded her - she could easily have been abandoned by history and been left in penury.

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 20:34 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link]

> I may be muddling up my stories, but I think it's this one, it IS the classic case of a lone researcher burning themselves out both emotionally and financially, and the vultures pile in once they realise it does actually have potential ...

The story is a _bit_ exaggerated: https://en.wikipedia.org/wiki/Katalin_Karik%C3%B3#Career

You also might note that without being able to patent stuff, her career would have been impossible.

The disabling of hardware codecs in community distributions

Posted Oct 18, 2022 14:49 UTC (Tue) by Wol (subscriber, #4433) [Link] (8 responses)

> I think we just don’t have terribly good answers to the *why* of patents that cover enough of technological activity, particularly software.

Because you're confusing mathematics, science, and technology.

Mathematics (software) is not Technology.

There is a philosophical argument as to whether Science is a subset of Mathematics (I say no), but both are part of Philosophy.

I said something like that elsewhere, but for me Mathematics is the building of models on a foundation of axioms. Proofs are proof of correctness. The only place I'm aware of that you can patent a proof of correctness is software - it would be the height of craziness if you allowed people to patent exam answers, but this is effectively what software patents do.

Likewise, Science is doing experiments. You take a mathematical model (theoretical physics, mathematics), and run experiments to see if reality actually agrees with the model (experimental physics, science). Proofs are proof that the model doesn't apply - that it's the wrong model. This isn't patentable either (or shouldn't be) - it's sweat of the brow, it's discoverable knowledge.

And lastly we have Technology. This could well be the tools we use to do our scientific experiments, but this is where we apply our maths and science to actually achieve something. Let's take the internal combustion engine - if you mix fuel and oxygen you do NOT end up with just work, CO2 and water. The presence of contaminants like nitrogen gives you NOx, the proportions of fuel and water control how much unburnt carbon is left, etc etc. Technology is taking your *desired* scientific result, and trying to achieve it as best you can in the real world. Patents are meant to protect how you achieve the scientific result. You should not be able to patent the theory, you should not be able to patent the result. Neither are technology, and neither are Patentable Subject Matter.

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 9:38 UTC (Wed) by paulj (subscriber, #341) [Link] (7 responses)

Science and mathematics are very different fields. Mathematics is an incredibly important tool in science, and there are sub-fields of science that are most entirely mathematics ("Theoretical X" typically ;) ), also maths is a search for truth as science is. However, science is not maths and maths is not science. :)

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 11:32 UTC (Wed) by Wol (subscriber, #4433) [Link] (6 responses)

As is obvious from my post, I agree with you.

However, I get the strong impression we are the minority. Both Science and Mathematics are proper subsets of Philosophy. Given the general belief that Philosophy == Mathematics, that makes Science a subset of Maths.

As I say, I don't agree with that, but ...

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 11:42 UTC (Wed) by amacater (subscriber, #790) [Link] (3 responses)

Wol: As ever, there's an xkcd for that https://xkcd.com/435 but I suspect we're now well beyond the topic.

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 11:48 UTC (Wed) by Wol (subscriber, #4433) [Link] (2 responses)

Well, as a physical chemist, I'm well pure :-) (No, I don't do any work there, never have, but that's where my education led me ...). I only acknowledge the existence of electron shells :-)

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 13:55 UTC (Wed) by hummassa (subscriber, #307) [Link] (1 responses)

If you put an electron shell to your ear, do you hear the sound of the quantum foam sea?

The disabling of hardware codecs in community distributions

Posted Oct 19, 2022 15:07 UTC (Wed) by Wol (subscriber, #4433) [Link]

:-)

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 13:05 UTC (Thu) by paulj (subscriber, #341) [Link] (1 responses)

I'd agree they (classically) fall under philosophy, sure. I would not say philosophy is maths though. Philosophy is the very broad area of systematic search for knowledge and truth, no?

The disabling of hardware codecs in community distributions

Posted Oct 22, 2022 12:20 UTC (Sat) by Wol (subscriber, #4433) [Link]

Philosophy is the use of logic to understand the world.

Maths is the use of logic to understand "the world".

It's a philosophical argument whether the two are the same ...

Bear in mind, "Science" didn't even exist until the 17th Century, while philosophy has been around since Aristotle and before. And Maths, we know, has been around since about the 10th century at least and, as I wrote above, the definition is arguably identical to philosophy ...

And it all probably hinges on the fact that you cannot use a system of logic (any system) to prove itself correct - if you do you end up with a proof that true equals false. So once you start arguing the distinction between Philosophy, Maths and Science, you end up arguing about how many angels can dance on the head of a pin. That's an extremely important philosophical question :-) Read Terry Pratchett, he has a wonderful explanation of why questions like that really do make logical sense ...

Cheers,
Wol

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 11:19 UTC (Sat) by beagnach (guest, #32987) [Link] (1 responses)

>> ...Or the parent company is based in a legal grey zone. *cough*

> Companies with sufficiently deep pockets that are not based in legally sheltered areas

Who or what are they hinting at here with the talk of some company based in a legally sheltered area?

The disabling of hardware codecs in community distributions

Posted Oct 21, 2022 6:26 UTC (Fri) by AdamW (subscriber, #48457) [Link]

It's a reference to Canonical's parent company being registered on the Isle of Man.

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 12:15 UTC (Sat) by ju3Ceemi (subscriber, #102464) [Link] (2 responses)

"Instead, the acceleration features in the hardware are viewed as a partial solution that only implements the patented algorithms when combined with software support on the CPU"

Yeah, well
The feature in software (Mesa) can be viewed as a partial solution that only implements the patented algorithms when combined with hardware support

As long as fedora and stuff do not provide my that hardware part, fedora lies on the same spot as those hardware manufacturers.

Legal people are sick people (or bored people, like security people, that enjoy fucking around with non-sense ?)

The disabling of hardware codecs in community distributions

Posted Oct 15, 2022 13:15 UTC (Sat) by nim-nim (subscriber, #34454) [Link] (1 responses)

Fedora wants to be distributed by some (most) of those manufacturers, getting pre-installed by manufacturers is the only way to make those interested in getting hardware quirks handled before users hit them.

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 13:05 UTC (Thu) by kpfleming (subscriber, #23250) [Link]

In which case those hardware manufacturers can arrange for an appropriate license to be included with the purchase of the hardware.

The disabling of hardware codecs in community distributions

Posted Oct 20, 2022 8:20 UTC (Thu) by hadess (subscriber, #24252) [Link]

As a quick work-around, Fedora and openSUSE users can install their video playing apps from Flathub instead, where the Mesa builds in the FreeDesktop runtime (and its descendants like the GNOME and KDE runtimes) don't have hardware codecs support disabled.


Copyright © 2022, Eklektix, Inc.
This article may be redistributed under the terms of the Creative Commons CC BY-SA 4.0 license
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds