Slashdot Mirror
Tridgell Recommends Reading Software Patents
H4x0r Jim Duggan writes "Andrew Tridgell rejected the common fears about triple damages: 'If you've got one lot of damages for patent infringement, what would happen to the project? It's dead. If it gets three lots of damages for patent infringement, what happens to the project? It's still dead.' Tridge then explains the right way to read a patent and build a legal defense: 'That first type of defence is really the one you want, it's called: non-infringement. And that is: "we don't do that. The patent says X, we don't do X, therefore go away, sue someone else, it's not relevant for us." That's the defence you want. [...] Next one, prior art: [...] Basically the argument is: somebody else did that before. It's a very, very tricky argument to get right. Extremely tricky, and it is the most common argument bandied about in the free software community. And if you see it in the primary defence against a patent, you should cringe because it is an extremely unsafe way of doing things.' There are even some tips in the talk specifically for Slashdotters."
In the same way that there is always a bigger fish.
And what you have to fear are overly broad patents and patent trolls.
Many patents are there not for suing the ass off the competition but to protect yourself from getting attacked.
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
When I'm writing code, I don't have to worry about patent infringement, because the stuff isn't patentable. That way I can write code just like you write anything else; without looking over your shoulder.
Why is the legal system so fucked up that it's easier to win with "We don't do X" instead of "X has been done before" when it's extremely obvious that X has been done before?
are along the lines of "Don't be casual and informal and creative with logic like Slashdot". He mentions Slashdot in the context of reading the abstract and not reading the actual thing, much like this very thread.
"we don't do that. The patent says X, we don't do X, therefore go away, sue someone else, it's not relevant for us."
"Look... me and the McDonald's people got this little misunderstanding. See, they're McDonald's... I'm McDowell's. They got the Golden Arches, mine is the Golden Arcs. They got the Big Mac, I got the Big Mick. We both got two all-beef patties, special sauce, lettuce, cheese, pickles and onions, but their buns have sesame seeds. My buns have no seeds."
The cost of searching for patents whenever you're doing something, anything, really, is a huge burden on any R&D department. By forbidding employees to look at patents, companies make then focus on the important stuff: making things.
To do list for Windows
I think that the best overall defense is if someone can take a case all the way to the supreme court, and argue that the threshold for obviousness
is way out of whack in the granting of software and business process patents.
The goal should be to get a test mandated by the supreme court along the lines of: If three out of four average 3rd year comp sci students
could design and code it up in a month, having heard only the requirements and not the design, then it's f***ing obvious to a competent practitioner
in the field, and should be dismissed.
Who knows. There may actually be algorithms clever enough to deserve patent protection for 5 years or so. But right now those would probably be
1 out of 100 of the patents granted, so the whole system is in total disrepute and it is the duty of a responsible practitioner in the field, as a protest,
to ignore and not view and thus not heed any USPTO software or business process patents, until they entirely revamp the standards and
throw out all patents granted under the existing broken standards.
Where are we going and why are we in a handbasket?
Can we just allow companies to patent everything they want and let them be effective for like only 5 years so we can all get on with life? Innovators should not be rewarded for how "clever" they are, but rather compensated for how much it costs to do R&D. Patent duration should depend on the costs associated with innovation in each industry.
I think more independent software inventors should do the same i.e publish the patentable details, as it will create submarine patent-eliminators against the patent trolls. Ultimately no one will know what really has been published, or if any software patent will hold in court. For example if I 5 years a from now will read about a patent troll suing someone and I know I have the prior art, well I'll be happy to be the expert witness invalidating the patent.
Are there not websites which accept submissions written to the standards required of a patent, for the very purpose of establishing prior art?
Has Tridgell patented, the techniques to defend patent attacks? Or is this going to be a prior-art for a later patent?
Yeah, taking it all the way to the Supremes worked really well in killing off infinite copyright extension, didn't it?
Reading patents with an eye to identifying "Technologies" to use is an exercise in futility:
Most are stupidly obvious. The others written in leagalease.
Reading patents with an eye to identifying "technologies" to avoid is also an excercise in futility. Again, you need the mind of a lawyer, combined with the approach of a security researcher, to "see" the ways a patent could be exploited to somehow map to your own problem domain. That you were happily solving without resorting to the giant database of solutions to micro problems no one is interested in.
Next, theres just too damned many of them. If anyone took the time out to exhaustively read and analyse each patent enough to determine if the possibility for collision existed, well they wouldn't have a problem with patents as they'd never write any code.
Lastly, it takes courts a long time to determine if a particular product does conflict with a patent. This means theres a lot of grey area around the edges of a patent to determine if a particular approach is covered or not. Which means, of necessity, that, like Chinese ISPs, developers who read a patent would have to defensively eliminate huge swathes of potential solution space from their investigations, to avoid getting "too damned close".
'If you've got one lot of damages for patent infringement, what would happen to the project? It's dead.
Not necessarily, since the product may already be released, and the project to create it may be already completed. The damages award is typically made after lengthy legal proceedings, and it's not likely that the development project is still active. Those involved have probably moved on to other projects, been promoted, or changed employer. Paying tens or hundreds of millions in damages is not fatal to large companies, but the risk of such a cost is definitely a risk to be minimized.
If it gets three lots of damages for patent infringement, what happens to the project? It's still dead.
Not necessarily, if the product has already been released, and the project to create it has already been completed. But the court's interpretation of a patent is often baffling to an engineer, who cannot therefore assess whether a risk of infringement really exists. However, by being aware of the patent, the potential cost of an infringement is tripled. This is why large corporations do not encourage their developers to do exhaustive patent searches.
Those who can make you believe absurdities can make you commit atrocities. - Voltaire
You could have at least read the summary:
Next one, prior art: [...] Basically the argument is: somebody else did that before. It's a very, very tricky argument to get right. Extremely tricky, and it is the most common argument bandied about in the free software community. And if you see it in the primary defence against a patent, you should cringe because it is an extremely unsafe way of doing things.'
The reason why prior art is difficult to get right is explained in TFA: a patent consists of a number of (likely very long and complicated) interdependent claims which are likely to be interpreted quite narrowly. In order to work, a prior art defence has to exist for each and every one of those claims in the same interpretation as that intended when the patent was granted. So you have to go through the entire patent from beginning to end, look at each claim and think "Is there prior art for this? Is the prior art exactly the same idea or is it just roughly the same sort of thing? If the latter, that's a Very Bad Thing for the defence".
However, it's quite common for the patent to hinge on a handful of claims and if you can prove that you don't do just one of the independent claims, you're free.
The union goes around aquiring as many patents up as possible, and agrees to use the patent portfolio to help defend any of its members who are threatened with patent legal action. All members agree to not to sue each other. Maybe a fee is required to join, or patent donations. Maybe a cross between insurance and a union. If it gets powerful enough, it could make patents irrelevant. ... or are people already trying to do this
And they cost about the same as filing a patent.
The Grey Goo disaster happened 3 billion years ago. This rock is covered in self replicating machines!
You could have at least read the summary:
I did, and I've had my share of these kinds of patent troubles. I no longer believe so much for (SW) patents and consider most of those waste of time. And I agree showing prior art is tricky, very tricky. And I hope, if needed, I've got enough both offensive and defensive patent skills if needed. In past I've done inventions, written patent applications and got patents. So I know the formal things. But as I publish the patent details, as you mentioned, using those interconnected patent claims, well chances are I publish something I *could* get a patent self as it described as a real invention. And if I manage to describe a patentable invention then it alone may serve as a defense - if their's isn't exactly like mine.
;)
...besides SW patents , who cares... I remember one brave attitude being shown with the in BBC's OS video codec project. They just said: if there would be patent infringement claims... they would just code around those....
And what about those offensive techniques, if you annoy me enough changes are I invalidate your other patents and hurt your R & D badly. Well I've done that too (in 2005-2006). Basically It took 2 weeks of HARD HARD HARD patent breaking activity, but it was enough to cause a lawsuit against me dropped - as in those two weeks I managed to make myself just too F*king poisonous to bite. But as said it was rough... very rough... but... it can be done. And no it was not a small entity that was suing me
So, I don't know what works for you... I just believe publishing SW inventions at the Usenet is an easy non-expensive way for making good quality prior art.
No, there are no algorithms clever enough to deserve patent protection. What is an algorithm in a computer? It is just math by another name. Software is made up of mathematical principles, math is not patentable, and therefore software shouldn't be, end of story.
I see several problems with Tridgell's approach.
1st, his talk assumes patents cover only a specific, narrow implementations of ideas. And if some patent troll challenges you, all you have to do is show your idea is slightly different. He says we've gotten it all wrong, because there isn't any such thing as an overly broad patent. I don't know. Isn't the point of a patent to cover an idea, no matter how it is implemented? What is a business method patent, if not that? But if he's right, then RIM really blew it. All they had to do was show that NTP's patents don't cover exactly, precisely what they did and the way they did it. That should have been easy, because they came up with their own system. Somehow, I think that if it had been that easy, RIM would have found the way. Instead RIM tried what he strongly recommends against, that is, they tried to show prior art. There was just this minor problem that their legal team foolishly hoked up some fakery on that point.
2nd, he assumes too much about the methods of the patent trolls. Sure, it all comes down to making money, but the straightforward approach of attempting to collect license fees isn't the only way to use a patent to that end. There's the use of patents to stifle competition. That's why MS supported SCO, not to profit off of licensing fees for Linux, but to hurt Window's biggest competitor. Then there's the shakedown, as IBM once did to Sun over 7 patents, including the infamous "fat lines" one. Sun did exactly what Tridgell says to do. They demolished IBM's claims of infringement. And it didn't work because IBM pointed out that they have over 10000 other patents. "Do you really want us to go back to Armonk and find 7 patents you do infringe?" Well, maybe Sun shouldn't have caved. And, there's harassment of the sort SCO did. They knew they didn't have a case-- they were just making a big legal stink in hopes their victims would decide it's cheaper to buy them off than fight them in court. How did PJ tear SCO's case apart? The way Tridgell recommends? Yes, but that was only a part of it. There was also effort to throw the validity of SCO's patents into question by showing prior art.
Intellectual Property is a monopolistic, selfish, and defective concept. It is "tyranny over the mind of man"
Of course prior art is a more popular argument in the free software community because it is a more aggressive defense. It's an attack against the patent itself. The free software movement is keen to destroy patents they consider unjust (which for broad and vague patents such as "taking online orders with a single click" or "online course management" is a given), and merely avoiding the damage from lawsuits may look like a "weak" move.
(Which is probably par for the course in law: Clients resent their lawyer's advice when they counsel against an aggressive course of action. Their rights were violated, dammit, and they want to see vindication. Even when the slower and less flashy strategy, even settlement, can have the better ultimate outcome.)
Claim 1) A [noun] consisting of:
* a [adjective] [noun],
* with [adjective] [noun] [noun]
You then use software to fill in the blanks with every possible word option:
Claim 1) A [computer] consisting of:
* a [grumpy] [fish],
* with [cloudy] [metal] [socks]
Run all of these permutations through, post them online with a time stamp and under the GPL code, et voila, the prior art of everything! Wait a year and a day and you can now claim every software patent is worthless because there is "prior art" - and all you have to do is reference the website that has the "prior art of everything." Hell, that should be the website: "priorartofeverything.com"
Then, whenever anyone files a software patent, point to "priorartofeverything.com" and say, "Nope. Someone already had that idea."
The end of software patents....
I'm sure Tridgell must know something about reading software patents. Take his PhD thesis for instance, which is "based" to a large extent on one of the IBM patents he references. Well I guess that one must fall into the first category he mentions.
The reason prior art is risky is because it's an affirmative defense that the defendant, not the plaintiff, bears the burden of proof for.
Back in the days of dial-up modems, Hayes tried to force Motorola to license the (pause)+++(pause) escape sequence [Heatherington].
The response was swift... Motorola looked in their own collection of patents to see what Hayes probably infringed....
To paraphrase, the response started with "You use wire, don't you?"
We’re actually supposed to read those?
But this is Slashdot. We don’t even read TFA.
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
I think he's wrong equivocating the invalidity defense with the prior art defense. My understanding is a patent can be invalidated - and rendered completely ineffective - if you can show that it doesn't actually teach a practicable implementation of a way to achieve the claims.
I had experience with this. We received a cease and desist letter from a (large) company saying we were infringing a patent they had claiming synchronizing audio playback with the movement of a cursor. After carefully reading the description, we realized that they were actually describing doing this synchronization by assuming that the real-time clock signal was all that you needed to know how much of the wave file had been sent to the audio output ... and we knew that this could not actually work. It didn't account for processing delays owing to CPU/memory/bandwidth limitations. Our lawyer wrote a letter back to them saying this and we never heard from them again.
Note that the _claims_ themselves did not describe the synchronization method - they were claiming the generality of doing the synchronization. It was in the _description_ that they explained _how_ to do the synchronization and this is where we found the flaw which invalidated the entire patent. I should note also that the description included words indicating that the method they were describing was "essential" to the invention - so it was actually a badly written patent. If they had carefully qualified the description with words like "this is one possible method ... there are others known to those skilled in the arts", we might not have been able to make this defense. And, of course, this never went to court (probably because they realized how badly the description had been written). But, I've seen other such flaws in patent descriptions - you'd be surprised how often lawyers make stupid mistakes like this.
And, if you do find such a mistake, you will have helped to move toward invalidating the entire patent, as opposed to just avoiding the particular infringement suit. It is lots more work to wade through entire descriptions, and I wouldn't recommend doing it unless, as the speaker indicates, you are in the cross-hairs of an infringement suit. But, it can be a very good feeling if you succeed!
If humans are mostly water, and beer is mostly water, then humans must be mostly beer.
So, would an infringer have to infringe on all of those claims as narrowly interpreted? If so, it seems like one could avoid patents by making some trivial changes to the code.
Cory Doctorow talking about cloud computing makes as much sense as George W Bush talking about electrical engineering.
Comment removed based on user account deletion
By "workaround" he didn't mean a technical solution where you implement something different to avoid infringment. He meant a legal solution where you look for an argument that says you're not infgringing.
Why is the legal system so fucked up that it's easier to win with "We don't do X" instead of "X has been done before" when it's extremely obvious that X has been done before?
Because "we don't do X" is a very easy, black and white conclusion. Whether something is obvious or not is a very difficult analysis, requiring examining multiple factors, the state of the art at the time of invention, the ease of combining different prior art references without undue experimentation, how skilled the average person in the profession is, etc. Compare it to other areas in the legal system - it's much easier to beat a murder rap if the victim is still alive, than to try to show that your questionable alibi is more persuasive than the questionable witness testimony.
that the best way to handle a problem is to fold around it?
And how many times are we supposed to do that before giving up, once the options are exhausted?
I think the best way to handle a problem is to remove it.
As nice as the world might be if food was free and software had no patents, that isn't the world we live in.
You may want to distinguish between the country you live in and "the world". Most european countries, for instance, still do not allow software patents at all.
Article 1, section 8, clause 8, reads, "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."
It does NOT read "to make money for the creators for themselves and their families in perpertua".
Furthermore, esp. in software, where something may easily be obsolescent in five years, to grant a patent that lasts 17 is to *not* promote the progress of science", but rather to restrict it.
But we've seen plenty of folks (gee, boyos and grllls, can you say 'SCO'? Or 'RIAA'?) who think it means the latter.
mark
The link in the Slashdot "article" isn't valid. No one seems to have noticed this. No, they just blither away in ignorance.
I believe it is the suggestion to execute particular sequences and combinations of mathematical/logical operations,
to achieve a particular purpose, which is being patented, not the abstract existence of the mathematical principles themselves.
That being said, I tend to agree with your "information shall be free" idealist position on these things, but then I also think
that clever / creative people should be given enough money to pursue their creative passions without starving or freezing to death.
But instead it's a dog eat dog world, so the clever but otherwise hapless dogs have invented patents to make
sure they don't starve if they produce useful inventions.
I guess they could just produce inventions that are badly designed and need lots of paid tech. support. Oh wait.
Where have I seen that business model before?
Where are we going and why are we in a handbasket?
I should clarify, "the suggestion to have a machine execute particular sequences/combinations of math/logic operations"
in order to achieve a "real purpose in the world" and not just achieve the calculation of the mathematical result.
Where are we going and why are we in a handbasket?
Seems to be working better with business processes patents (we'll be sure on september). Maybe somebody should put software ones before the court.
Rethinking email
One can avoid most patents by making trivial changes to the code/product.
Rethinking email
Software patents are evil. Accepting them and trying to circunvent them is not going to help. You cannot circunvent what you cannot be aware of (hundreds of thousands of software patents waiting to bite you as soon as you have enough money for them) More or less one year ago I started the petition: http://petition.stopsoftwarepatents.eu/ More than 50.000 people and 3.000 companies have signed, but many haven't. I have spoken with some in the later group and their reasons are very different from each other. And that is the problem to stop software patents: many think approach a) or b) are not a good idea (or are useless). But that is wrong. All approaches against software patents must be supported (specially if they don't cost you too much). Go and sign the petition now ;-) (even if you are non-european)
Let p' be "not p". What's the difference between showing "not p'" and "not q"? Why can't the patent owner say "We'll try to show Not-p'" rather than "[...] Not-Q"? Why don't they say "OK, P-And-Q so we can't enforce patent"?
You say something about legal wrangling, without any details. I don't think you can argue that one works better than the other based solely based on propositional (or if you prefer, second order prepositional) logic---at least not the kind of logic that deals with what is true rather than what is convincing.
I think you need to argue from the realities of court rooms and court cases: that based on historical evidence we know that one kind of argument convinces people and another kind of argument does so too but less often. Maybe you can argue based on what the law says: that "it has been done before" isn't enough to constitute prior art, and therefore Q is a necessary but not sufficient condition for defending yourself (that makes your second scenario, juxtaposed with your first, make sense).
But calling the claims one-letter names and formalizing the structure of the argument, what exactly has that gained you? Exactly what are you claiming here?