LWN: Comments on "VFAT patent avoidance and patent workarounds"

VFAT patent avoidance and patent workarounds

nix — Wed, 08 Jul 2009 18:16:58 +0000

Well, we've already seen "don't discuss details of the implementation of
this patch: $PERSON will send you a private email explaining the design"
(and then he apparently doesn't, at least not yet) which means that if
this patch goes in we have opaque design decisions in the kernel for the
sake of a single (large) country's appalling legal regime. Great stuff.

I wonder what can't-discuss-it code will be going in for the Chinese
government next? They're hot on Linux and China has a much bigger
population than the US.

VFAT patent avoidance and patent workarounds

fergal — Wed, 08 Jul 2009 11:17:31 +0000

I don't see how it wouldn't crimp our style discussing anything to do with the scheduler forevermore. How are we going to discuss LFN issues of any sort in VFAT in the open from now on? As far as I can tell, we pretty much can't.

You can discuss almost everything. Just don't publicly say "hey I don't think this does work around the patent because ...". If you think you've found a flaw in the legal reasoning, send a private email. You probably also shouldn't send a patch for purely legal reasons and discuss them in public.

That leaves everyone who is likely to do any work on LFN reasonably free to do it and discuss it in public.

VFAT patent avoidance and patent workarounds

giraffedata — Fri, 03 Jul 2009 19:03:45 +0000

Code is a set of instructions performed by a machine.

That's a bad way of thinking about code. It's an accurate description of machine code, but since the work of Grace Hopper, good code is a description of the solution to a computational problem. It's written in a language simple enough that a computer can understand the solution and implement it. So calling code instructions to a computer is like calling a manuscript instructions to a publisher.

So the way I would compare law to code is that law is what code would be like if we had much more advanced computer technology (and I do think we'll get there eventually). Imagine code that technically says to iterate forever, but the computer understands you didn't really want an infinite loop and stops after a reasonable number of iterations. Law can do that.

US constitution vs patents

njs — Fri, 03 Jul 2009 03:19:36 +0000

I believe you're thinking of Eldred v. Ashcroft. The US Congress certainly *does* need authorization from the Constitution to pass laws; that's one of the basic principles of US law. (The Constitution basically says "here's a list of things Congress can pass laws about, here's a list of things that no-one can pass laws about (regulating speech, etc.), everything not mentioned here the states can do whatever they want with.")

IIRC, the Supreme Court wriggled out of the Constitution's language about copyright being "to promote the progress of science and useful arts" by saying that it was complicated to say exactly whether progress was hurt or helped by copyright extension, and so they would leave that up to the legislature, who has more resources to figure out tricky questions of this sort. Kind of lame.

VFAT patent avoidance and patent workarounds

mjr — Thu, 02 Jul 2009 14:45:22 +0000

Well, I certainly hope this improved patch will go in, as pragmatism in an imperfect world would dictate.

Regardless, the patch is pretty simple, so at least vendors can apply it on their systems without too much trouble and maintenance effort to avoid becoming collateral damage in Microsoft's FUD war.

VFAT patent avoidance and patent workarounds

grahammm — Thu, 02 Jul 2009 05:47:53 +0000

And that shows that the current advice that a practitioner should never look at patents also goes against the intent of patents. This shows that patents are intended to be a mechanism for sharing knowledge amongst practitioners. If practitioners do not read patents, or if they are written in 'lawyer speak' rather than language understood by the practitioners, then they are not serving to "promote the progress..".

VFAT patent avoidance and patent workarounds

faramir — Thu, 02 Jul 2009 04:39:14 +0000

It seems to me that this is a similar situation to people signing NDAs with hardware manufacturers and then writing free software drivers. There are some things they aren't telling you and will refuse to discuss with you.

Why people are upset about a similar situation involving patents, I just don't understand.

VFAT patent avoidance and patent workarounds

njs — Wed, 01 Jul 2009 22:47:07 +0000

As part of the rising interest in "green" computing, I propose that henceforth we make analogies involving biking, walking, and light rail whenever possible.

VFAT patent avoidance and patent workarounds

nix — Wed, 01 Jul 2009 22:05:53 +0000

That works, too.

Anyway, my apologies for introducing an analogy unrelated to cars, but as
I don't drive I have to make my own fun.

effect of discussion on legal procedings

pjm — Wed, 01 Jul 2009 13:46:47 +0000

I haven't seen anyone say that whether or not one says things changes whether the code infringes the patent. The FAQ says that it can change whether or not the judge can give a summary judgement that there's no infringement: it influences the decision of whether or not there's anything that needs to be looked at more carefully to decide the matter. The FAQ says that judging on all the relevant facts is time-consuming and expensive; and that the aim is to avoid there being enough doubt that the judge decides it's necessary to do that.

Someone else in these LWN comments also suggested (without citing any evidence) that such discussions can change whether or not one is found to have knowingly ("willfully") infringed the patent.

effect on discussion

pjm — Wed, 01 Jul 2009 13:26:50 +0000

Just to clarify, I haven't seen anyone claim that we can't discuss the *implementation* of long filenames (or scheduling or whatever other hypothetical target of patent suits), just that it's dangerous to try to discuss what things would or wouldn't infringe the patent.

US constitution vs patents

pjm — Wed, 01 Jul 2009 13:07:52 +0000

A couple of years ago, people made this sort of argument in challenge to the extension of copyright duration, arguing that it doesn't promote progress (especially the extension of copyright of works by already-dead authors). The judge ruled that just because the constitution doesn't explicitly allow a law doesn't mean that legislators can't pass that law.

(Sorry I have no references. No doubt I'm not exactly reproducing the arguments on either side.)

VFAT patent avoidance and patent workarounds

njs — Wed, 01 Jul 2009 05:27:31 +0000

They can tell you why they wrote it. They wrote a whole FAQ. They're just asking that if you disagree with their (public) reasoning, you attempt to hash things out in private first, because in this case, paradoxically, a public discussion of a *potential* flaw could, itself, become an *actual* flaw.

VFAT patent avoidance and patent workarounds

njs — Wed, 01 Jul 2009 05:24:38 +0000

A successful summary judgment may prevent the trial ever reaching discovery, though.

VFAT patent avoidance and patent workarounds

njs — Wed, 01 Jul 2009 05:22:44 +0000

> That's why I mentioned old laws getting ignored: that's one major difference, because old code doesn't get ignored

Heh, I just read it as part of the analogy, with the old laws like the old modules that end up sitting in a corner mumbling quietly to themselves without actually being hooked into anything that matters anymore.

VFAT patent avoidance and patent workarounds

sbergman27 — Wed, 01 Jul 2009 00:54:19 +0000

Are you talking to me, or to the poster I was replying to?

I have found the discussion in this thread to be interesting. I would not necessarily have expected it to go in the direction it has. Sometimes, on this site, an essentially political story spawns technical discussion which quickly goes over my head. Other times a technical article results in a political discussion. It's really hard to predict. But few sites on the Internet consistently have such *good* discussions. And few flame wars, despite the fact that we are an opinionated bunch.

I suggest that you might want to reconsider your opinion on allowing comments to patent-related articles. I'm not at all sure that cloaks and daggers are likely to benefit our community.

VFAT patent avoidance and patent workarounds

gdt — Wed, 01 Jul 2009 00:27:48 +0000

Yes, your point on the futility of the legal wisdom of questioning the effectiveness of the patch is well made. But making a point isn't really what comments on LWN are about. Rather comments are a service to the reader, to point out errors in the article or alternative points of view. Should Microsoft decide to sue a party over the VFAT patents I'm sure the pain in the butt that is discovery and deposition will gain us readers a measure of revenge for your rather poor joke.

On a related note, I've no idea why Jon permits comments on these patent patch articles. It is painfully obvious reading the comments to date that comments are only hurting the effectiveness of those patches, rather than contributing to a discussion of a high enough quality where upholding 'free speech' might be worth the consequences.

VFAT patent avoidance and patent workarounds

sbergman27 — Tue, 30 Jun 2009 23:56:28 +0000

"""
(I am one of the best VFAT experts in the world)
"""

My sincerest condolences.

VFAT patent avoidance and patent workarounds

nix — Tue, 30 Jun 2009 23:01:55 +0000

I don't see how it wouldn't crimp our style discussing anything to do with
the scheduler forevermore. How are we going to discuss LFN issues of any
sort in VFAT in the open from now on? As far as I can tell, we pretty much
can't.

(A more significant reason why nobody reads patents is because they're
written in such an appallingly unreadable turgid style that by the time
you've figured out what they're talking about, you could have come up with
the idea yourself in 99.9999...% of cases. This in itself is an indictment
of patent quality... a patent library that nobody consults is entirely
worthless --- except as a weapon to use against the public. But you knew
that.)

VFAT patent avoidance and patent workarounds

jzbiciak — Tue, 30 Jun 2009 22:17:44 +0000

This is why so many technical folks don't proactively go read the active patents at the USPTO (or other patent organization). You can't willfully infringe if you are truly unaware of the patent.

For example, if this VFAT patent was just buried in the piles of MS patents and nobody had brought it up, we wouldn't have this patch. It's not until MS starts flexing its muscles that we say "ok, what now?"

If some patent troll comes along claiming against our scheduler, there's no claim of willful infringement at that point, because we simply didn't know about it. Discussing the workaround gets tricky, and that's the part that sucks. Invalidating the patent with prior art also is interesting at that point.

In any case, in your imagined scenario, the mere fact of the patent's existence didn't stop us from coming up with our version of the scheduler to begin with. It just crimps our style in discussing how to handle the patent troll once he shows up and only if he shows up.

It sucks and I certainly don't like it. There may be some chilling effect, and if there is, evidence of it should be held up as shiny examples of why the status quo is bad for innovation.

VFAT patent avoidance and patent workarounds

nix — Tue, 30 Jun 2009 21:54:36 +0000

I was actually being cynical about code quality, rather than reductionist
about the law. I suspect that code *will* get as hopelessly inconsistent
as legal systems do, if you let them accrete for as long: i.e., the
driving factor here is time.

(That's why I mentioned old laws getting ignored: that's one major
difference, because old code doesn't get ignored. Machines have no common
sense. Most judges do, and even some legislators...)

VFAT patent avoidance and patent workarounds

jordanb — Tue, 30 Jun 2009 21:29:03 +0000

> I'd say the law is *exactly* like what code would be like if it was
> written over many centuries, by people who never understood more than a
> fraction of the codebase, and with a goal of consistency above all despite
> that.

I think that's an excessively reductionist view of human society and the way it interacts with its laws.

Code is a set of instructions performed by a machine. This is true even of 'big ball of mud' type systems. Any attempt to reduce the operation of human society to that of a machine will get into the more extensional (and pointless) aspects of the philosophy of AI.

VFAT patent avoidance and patent workarounds

marcH — Tue, 30 Jun 2009 21:19:14 +0000

This workaround is not effective. Linux is still infringing the VFAT patent even after applying it.

(I am one of the best VFAT experts in the world)

"obviously duplicated" is irrelevant

nix — Tue, 30 Jun 2009 21:06:18 +0000

Have you *seen* VFAT? If an undergraduate student implemented something
that ugly he'd get a C, max.

(Of course, nobody wanted it to be so ugly. VFAT is the way it is because
of back-compatibility with a bunch of now-dead systems...)

VFAT patent avoidance and patent workarounds

nix — Tue, 30 Jun 2009 21:03:16 +0000

I'd say the law is *exactly* like what code would be like if it was
written over many centuries, by people who never understood more than a
fraction of the codebase, and with a goal of consistency above all despite
that.

(In many sufficiently long-lived legal systems, the very oldest laws don't
even get repealed: they just drop out of use because almost nobody can
even understand the language they're written in, and they're not important
enough for the legislature to waste any time on as nobody ever actually
uses said moribund laws for anything. Some of the oldest
still-technically-valid UK laws are written in Anglo-Norman, but unless
they've been renewed more recently, they're probably forgotten except as
curiosities. Most laws that are actually enforced are either terribly
significant so they don't get forgotten or a couple of centuries or less
old. This probably helps keep the set of active laws from imploding from
sheer size...)

VFAT patent avoidance and patent workarounds

nix — Tue, 30 Jun 2009 20:57:44 +0000

So what happens if (when, more likely) someone claims a patent on some
aspect of task scheduling or memory management or something else core to
the OS? Must we stop discussing *that* on l-k as well?

This alone is quite capable of destroying open software development. They
don't need valid patents: this chilling effect on its own will do it.

(And now they've seen the effect of asserting a patent on something
relatively minor which they unquestionably did invent, like VFAT, who
knows what'll happen next?)

VFAT patent avoidance and patent workarounds

jordanb — Tue, 30 Jun 2009 19:09:32 +0000

When did being knowledgeable or informed become a prerequisite for having an opinion on the internet?

"obviously duplicated" is irrelevant

ianw — Tue, 30 Jun 2009 18:28:23 +0000

I realise the issue is not duplication here, but how could a reasonable person not just roll their eyes and sigh if two lawyers were standing in-front of you duking it out over how completely different their VFAT implementations are (or not) because it slips though a loophole that somehow escaped the prolix that is a software patent.

I completely blame Microsoft, however. By forcing the TomTom issue they showed intent. If the best they can come up with is to sue over a file format vaguely more complicated than what we expect undergraduate students to implement to pass a basic operating systems course, then they are justifiably doomed.

VFAT patent avoidance and patent workarounds

clugstj — Tue, 30 Jun 2009 17:32:40 +0000

I was hoping that someone would respond to my comment's content, but let me try again.

My point is that this kind of patch (we can't tell you why we wrote it, just trust us), should be rejected on those grounds alone. Secrecy is deadly to the openness upon which Free/Open Source Software is based.

I also think the lawyers are covering their own backsides and laying a minefield for everyone else with wording that draws attention to the possible patent issues that they added in the configuration flag that enables the current behaviour (read/write both short and long filenames).

VFAT patent avoidance and patent workarounds

drag — Tue, 30 Jun 2009 17:09:28 +0000

It's not up to you to make those sort of decisions wheither or not patents in their current form are constitutional or not.

It's up to the supreme court and they ruled some years ago in a manner that is interpreted as allowing software patents.

Currently in order to get something declared unconsitutional you'd have to get into a patent case and then fight your way up to the supreme court and get them to agree with you.

If you want to go that route I hope that you have:

1. About 5 years or so.
2. Best lawyers in the country
3. Several million dollars burning a hole in your pocket.

Because that's about what it would take to even have a ghost of a chance to win a case like that.

And everybody and their mom will be fighting you. For two reasons:

* There is still doubt that pure software patents are 100% legal.
* There are a lot of very good lawyers that make very good money on software patent BS.

So if you....
Win: You put a large number of very wealthy and experienced legal people out of work.

Lose: You essentially destroy any chance of fighting software patents until you get Congress to pass a law making software patents illegal.

VFAT patent avoidance and patent workarounds

drag — Tue, 30 Jun 2009 16:59:48 +0000

No. If your arguing that a patch doesn't work around a patent then essentially your admiting that you've read the implimentation and that you know your currently violating it.

That's the only way that a judge is going to interpret that.

It's just logical. How is somebody going to 'poke holes' in a patent-workaround when they don't even know the patent itself?

VFAT patent avoidance and patent workarounds

mjthayer — Tue, 30 Jun 2009 16:25:30 +0000

Can that sort of thing be taken to the constitutional court in the US? Probably not if no one has done it yet...

VFAT patent avoidance and patent workarounds

dgm — Tue, 30 Jun 2009 15:03:12 +0000

> By questioning the patent-workaround to much your admitting that your KNOWINGLY violating the patent.

Maybe. Or maybe not. I would say that the only thing you admit is that you know that the patent exists. Maybe the questioner doesn't understand the implementation, or the patent, or both. Maybe the patent is bogus, maybe the actual code does not infringe (aren't patents open to interpretation?), maybe...

> I am not a lawyer. I haven't looked at the code, I haven't looked at the patents in question. etc etc.

Me neither.

The law is/is not code

davecb — Tue, 30 Jun 2009 14:37:03 +0000

The law is not code, it just uses some few of the same terms in ambiguous ways, such as "code". It's the black art of coding for autonomous processes (humans) using only global variables and universal statements like

for all X, if X is blue, do not do X

Arguing about whether X is in fact blue, having publiclly said we're doing X, is evidence that there is doubt about whether X is blue.

In doubtful cases, a company can ask a court to rule on whether X is blue, and if so, whether we've roken the rule.

--dave

VFAT patent avoidance and patent workarounds

drag — Tue, 30 Jun 2009 13:30:04 +0000

The #1 thing to remember is:

THE LAW IS NOT CODE.

The Law is not code. It simply does not operate on the same level of exactness that source code operates.

There are all sorts of side rules, all sorts of implying and interpretation, and all sorts of crap like that. This patent IS A PROBLEM for some Linux users and developers. The FAT is a very real, very expensive thing to deal with. Ignoring it and minimizing it is simply doing absoletely nobody any favors.

It's very difficult for programmers to understand that Law doesn't make sense. It doesn't have to make sense, it'll never make sense. It's something that is intentionally vague in order to be able to deal with the complexities of human interactions in society.

So it's entirely up to a Judge to decide on a case by case basis.

So it's entirely and 100% makes sense that doing things like having Linux kernel developers questioning the patent workaround is the same of the Linux kernel developers saying that they think that they are violating the patents.

Get it?

By questioning the patent-workaround to much your admitting that your KNOWINGLY violating the patent. At least it will seem that way, which is _good_enough_. It's opening up a huge hole in any sort of legal standing you may have.

It's not very complicated here.

I am not a lawyer. I haven't looked at the code, I haven't looked at the patents in question. etc etc.

VFAT patent avoidance and patent workarounds

bokr — Tue, 30 Jun 2009 08:12:39 +0000

Perhaps it could be fruitful to recall the origin of the
authority for Congress to create patent and copyright law.
It's only a few lines in the Constitution:

Article. I. Section. 8. The Congress shall have Power
[... many "To ..." items elided] ...
To Promote the Progress of Science and useful Arts, by
securing for limited Times to Authors and Inventors the
exclusive Right to their respective Writings and Discoveries; ...

That's all. Clearly (to me), this was not intended to allow
Congress to create laws that license neo-luddites to abuse
said laws to impede the "Progress of Science
and useful Arts" for any reason -- perhaps least of all
as a means of hobbling competition with their own outdated
and/or inferior technology.

Nor, it seems clear to me, was this provision of the Constitution
meant to permit or nurture the existence of patent-trolls, who if
anything impede rather than "Promote the Progress of Science and
useful Arts," and who are generally neither Authors or Inventors.

Exclusive Right can not be interpreted as a right to enforced
dis-use and be consistent with Promotion of Progress!

I am against software patents, but a practical way to "Promote
the Progress of Science and the useful Arts" might be to provide
legal immunity against patent infringement claims for non-profits
and academic and research entities, as well as entrepreneurial
businesses grossing less than say a million dollars.

After some product containing a patented entity is making real
money in the market, then the first to achieve this could have special
rights to a cap on royalties, and others would have to negotiate with
the patent owner or stay under the million gross limit.

This would provide incentives Promoting Progress through competition.
And FLOSS could promote progress as it does now, through cooperation.

VFAT patent avoidance and patent workarounds

jzbiciak — Tue, 30 Jun 2009 07:27:59 +0000

I think there's an important detail being missed:

The FAQ makes an important point related to workarounds that the community should hear: publicly questioning the effectiveness of a workaround can have fatal results.

It's not that we can't describe exactly what we're doing. (The description of the VFAT patch seems pretty complete here, with all its rigarmarole to avoid automatically writing two filenames for the same file to the directory structure.)

What's problematic is the process of poking holes in the workaround in public. Even if the "hole" you poke is imagined, creating the impression that the implementers know that their workaround doesn't actually work around the patent gets you back into the appearance of "willful infringement" territory. At least, it becomes easier to make that argument in court, particularly if the court isn't as tech savvy as the implementers are.

Keeping the hole-poking discussions private is what's more important. Yes, it's a loss of transparency, and yes, we should hold this up as a shining banner of why software patents are bad.

VFAT patent avoidance and patent workarounds

spiv — Tue, 30 Jun 2009 05:09:39 +0000

"Somehow by not telling you things, it changes whether our code would be found to infringe a patent"

Actually, the FAQ referring to the legal advice doesn't say that; it says that if someone said on LKML that they thought the patch might not avoid the patent then it makes it more likely that the case would go to a full trial. It doesn't say that it is likely to affect the end result of a full trial. But it does make the point that a full trial is going to be much much more expensive (and time consuming), so it's in the interests of a potential defendant to avoid it.

In short, the purpose of the patch is to reduce risk of litigation even starting, and a hypothetical public statement on LKML that it might not avoid the patent has the potential to at least partially undo that.

(IANAL etc.)

VFAT patent avoidance and patent workarounds

jamesh — Tue, 30 Jun 2009 04:31:29 +0000

If you think about it, it makes sense for Tridge to be working on this. As well as working around the patents the patch needs to maintain maximum compatibility with Windows systems, which is something he has a lot of experience with.

VFAT patent avoidance and patent workarounds

jlokier — Tue, 30 Jun 2009 03:37:39 +0000

"carefully examining the patent and tweaking your project so that it doesn't infringe is perfectly legitimate".

Quite, and that's exactly what the patches do. And everyone knows it. So why the "we can't tell you exactly what we're doing" part?

One could get the impression that even if it's legitimate to do, it's not legitimate to admit to doing it. Which seems thoroughly perverse.