https://twitter.com/vranieri/status/647179711563431940
I've seen this before (on BOTH sidees, to be clear): someone changes a plain meaning of a term to avoid losing. Not sure that's what's happening here, but it looks like it.
edit to add: Because of some confusion about the purpose of this post. It is to point out the flaws with the current system.
Words in patents are twisted "like a nose of wax" in order to arrive at a particular outcome (and as I mentioned, both patent owners and alleged infringers do this on occasion). A patent system that allows this then completely undermines the public notice function of patents. A patent should tell the public not only what the patent owner claims to own, but also what is free for others to use.
That failed here. Someone who wants to avoid infringement, on reading this patent, can't tell what they can and cannot do.
In the end, parties are spending thousands (likely millions) of dollars to figure out what "integer" means. Why? Because the patent owner, who was in the best position to tell the world what she invented, used a word that has a very precise meaning when other words could have avoided all of this.
In any case, the EFF loses credibility when it overstates the position to make someone look more ridiculous than is warranted.
The EFF does great work, but they need to dial the clickbait-o-meter down a bit.
As someone who has ordinary skill in mathematics, (and cperciva above, who has extraordinary skill in mathematics, agrees), I see the use of "integer multiple" here not as a unambigious plain meaning at all. It's akin to saying "This patent covers the use of lasers in blah blah" -- did the patent intend to include a case where exactly one laser is used? The answer depends on context -- real lawyering, not language-lawyering.
But where the issue comes up is if that's not clear. Patents are supposed to put people on notice of what is--and conversely, what is not--free for others to use. Here, by using the word "integer", which has a very precise and definite meaning in math and science, that conveys certain information.
Part of the problem with the current state of our patent laws is that claims are not clear and the public isn't on notice of what the patent owners are claiming. This discussion is a good example of that. If the patent owner was clear in what they meant, would we be having this discussion. Would the patent owner itself have claimed that "n=1" infringes if it was clear that n>=2?
Mathematics precisely defines the term "integer." Mathematics also provides quick and easy ways to exclude "1". None of that was done here, to the detriment of the public's notice of when they were infringing.
Patents, despite how esoteric they have become, are ostensibly written for those with skill in the art. If we're all having problems understanding what, exactly, the patent owner meant (including the patent owner!) there's a problem.
*edited for clarity
They can do that in this case if their only use of the word "integer" in the entire spec is in reference to this parameter, and if it has been otherwise clear all along (at least "to someone skilled in the art") that it doesn't make sense for the parameter to be less than two.
If I write a spec in which some number occurs that can only be 1, 2, 3 or 4, and it is the only integer, and the constraint to those four values is clear all along, there is no harm with later adding a glossary item which says "integer: in this specification, a noun denoting one of the values 1, 2, 3 or 4". It just reflects the fact: the only use of integer in the spec is in reference to a value which is one of these four.
If the 3 case infringes on some other patent, but the 1, 2 and 4 cases do not, what's wrong with reducing the scope of the patent by removing 3, and adjusting the glossary item?
Of course, patent holders shouldn't have infinite latitude in remotely manipulating the meaning of the text at will by means of shifting glossary entries.
When are you adding this? If you do that when obtaining the patent, all is well and good.
If you do that in court, after saying otherwise, as has been done here, there's quite a lot of harm. People have a right to know what is and is not patent infringement, as they're on the hook for big money.
If you can just change what your patent does and does not cover, on a whim, when fishing for money in court, there's quite a lot of harm in that. Millions of dollars worth of harm. Nobody had any way of knowing what idiosyncratic definitions of words you were using and yet everybody else was responsible for not infringing upon your patent.
But CW's current filing contains this:
According to the patent, “The MAC-e PDU is sent to the physical layer every n*TTI, instead of once every transmission time interval (TTI).” If the integer were to be 1, the MAC-e PDU would be sent to the physical layer once every TTI, which is expressly excluded by the specification.”
In other words, the specification excludes the n=1 case, and the n<1 cases are nonsensical.
The part that really bothers me in all this is that, reading the EFF article, I was wondering why they kept going on about how absurd it is to argue that 1 isn't an integer, but then never specified how the patent trolls came to make that claim. It's now pretty clear to me: because if they had explained it, people might actually have thought it to be a reasonable argument.
I held the EFF in pretty high regard before. After this? I'm not so sure any more.
N=1 is hardly an unreasonable multiple to choose for a time interval. If they meant something other than the mathematical definition, they should have properly defined it.
We would never say that the compiler should have known from context that I obviously meant unsigned int when I wrote signed int. Patents are also written in formal language.
No one should have to guess about what they might or might not cover when there are huge financial penalties for being wrong, or even being close enough to being wrong to wind up in court.
Then they decided that it obviously was not covered by their patent.
I don't think the courts should play this game any more than your compiler should simply decide it can figure out what you really meant whenever your program contains an error.
Words used in patent claims are always defined by how they are used in the context of the patent description.
You can't read the claims and think you understand what they cover, without reading the description. A thesaurus doesn't matter, wikipedia doesn't matter, the opinion of an math professor doesn't matter. Patent description.
Even the patent holder said N=1 was obviously included... until that presented a problem and they decided it was obviously not what they intended.
Why is it the case that people believe that the public, who did not write this patent, should have to guess regarding what it may or may not cover when they're also at threat of millions of dollars in penalties should they guess wrong? And that's neglecting court costs & attorney's fees, which are almost always a sunk cost--you pay them merely for getting sued and you have essentially no chance of recovering any of that, even if you're right, unless they essentially get laughed out of court because it's your burden to prove that the case was exceptionally bad... even though you did nothing wrong.
The original patent used "integer" in a place where the mathematician's Z can't possibly be what they meant, and the dispute is over whether the patent covers the "n=1" case or only the "n>1" case.
From the legal files:
> Here, the “integer multiple of transmission time interval” (or “n times TTI”) describes the interval of time between subsequent new transmissions of packets (called “MAC-e PDUs” in the patent).
> An interval of time between transmissions cannot be negative (which would nonsensically put the subsequent transmission in the past);
> nor can it be zero (which would nonsensically make the transmissions occur simultaneously and instantaneously).
> According to the patent, “The MAC-e PDU is sent to the physical layer every n times TTI, instead of once every transmission time interval (TTI).”
> If the integer were to be 1, the MAC-e PDU would be sent to the physical layer once every TTI, which is expressly excluded by the specification.”