SoylentNews is people
SoylentNews
For more than three years now, Microsoft has held to the line that it has loads of patents that are infringed by Google's Android operating system. "Licensing is the solution," wrote the company's head IP honcho in 2011, explaining Microsoft's decision to sue Barnes & Noble's Android-powered Nook reader.
Microsoft has revealed a few of those patents since as it has unleashed litigation against Android device makers. But for the most part, they've remained secret. That's led to a kind of parlor game where industry observers have speculated about what patents Microsoft might be holding over Android.
That long guessing game is now over. A list of hundreds of patents that Microsoft believes entitle it to royalties over Android phones, and perhaps smartphones in general, has been published on a Chinese language website.
More details are in the story, but too much to include in this summary.
(Score: 1) by cafebabe on Tuesday June 17 2014, @06:41AM
Patents are supposed to be written like scientific papers so that experts can learn from them and we (presumably) all benefit from them in the long-term. However, most patents are deliberately obtuse and some avoid using any meaningful terminology. A full-text search on a corpus of patents is an exercise in serendipity and critical patents can hide in plain sight.
1702845791×2
(Score: 3, Interesting) by Anonymous Coward on Tuesday June 17 2014, @09:09AM
As someone currently fighting to get a patent accepted for my company (despite actually hating the damn things).
I did *too* good a job in the description. It ended up looking so obvious (it isn't, literally rocket science applies) that the examiner rejected it on that basis.
(he's actually wrong: It's really not so simple.)
Now, if I was sure the rejection was valid enough to stop anyone else trying to apply for it (with a more obtusely written application that doesn't "look obvious"), I'd be quite happy to just let it go. My boss would be cool with that as well; because even failed, it becomes automatically searchable prior art itself.
And we do developement, so we just need to be free to do our job.
Anyway, it got me thinking: To be applicable, patents have be "not obvious to one versed in the art".
But in fact, this is a problem: Every useful patent describing something that actually works is necessarily obvious in light of its disclosure. It must work, therefore if you understand the applicable science, you can engineer it so it does.
But if written up so well as to be approachable by a layperson - which as you say patents *should* be: Then an average, or even slightly above average patent examiner is more likely to reject it because he/she finds it to obvious, compared to other patents they examine.
This would be because: There might be some (in the extreme case) submitted by, say, a less-than-average intelligence person who (mistakenly) thinks they have The Next Big Thing...
They will throw in a patent so convoluted that it would take an extremely highly intelligent examiner to realize that it actually wouldn't work: Most are going to think: "Yeah, looks non obvious, sounds feasible, certainly novel, guess I'll grant it."
Really, the apparent complexity is due to confusion / lack of clarity in the mind of applicant...
But then this patent ends up only wasting time and money because it was never a useful idea in the first place. Patent's aren't actually checked for physical validity - and it follows that such exist, and have negative global community value. They are noise.
Worse, in order to meet the "rules", the claims must be written in legalese: Essentially a Programming language for humans. It's as obtuse to read as any machine language, because like machine languages - it's definitions are singular, and it's logic precisely correct. (Ie, not a "human" language at all, despite the apparent similarity to conversational english).
This of course means that only lawyers can use it, making patents of fairly little use to anyone who might otherwise benefit: The "good ones" are good patents, but poor guidance. They also spend no effort to disclose hazards and traps, which is a pity, because most of the substance of wisdom is all about what doesn't work, and seems like it should.
I also read something recently about robust programming: the idea was that any complex program can and ought to be broken down into progressively smaller parts, until each part is obviously correct, and so the whole becomes obviously correct. (at the cost of the whole system potentially being harder for a newcomer to grok because of the larger number of abstractions, but I digress.)
Same idea can and should be applied to any working design. On the basis that any mechanism which does work, must effectively do so in the real world's "native execution environment", so to speak.
So then, no patent can be valid, because it follows that any idea, no matter how seemingly complex, necessarily can be reduced to a finite number of simple, obvious components.
And so it just takes a sufficiently intelligent person, with a big enough mental reach to grasp the entire "complicated idea" in all it's complexity, to re-encode it in sufficiently simple language so as to make it obvious.
The think about simplification by encapsulation is, everything becomes a large number of black boxes, and a mistake in meaning concerning any one of the types is enough to obscure the meaning of the whole. This is why OO languages have a harder time getting large communities around them: The languages themselves effectively become "programmable", and so every with that level of flexibility becomes in effect it's own dialect of the language, to the point that a newcomer must relearn the vocabulary specific to that application before they can even grok the code. Good think linux is written in C, which despite it's worts/traps, isn't so flexible: It's "smaller" to learn, and so one knowing C can open any one file in linux, and stand fairly good odds on figuring it out.
Simplification by encapsulation, though, means in essence zooming out: But if you can see the whole forest, you might miss some small, critical detail because your vision lacks sufficient resolution. So it is with any complex system/idea / pattern: Those who are "intelligent" merely have the talent of seeing not only the big picture, but in sufficient detail to have a chance at picking the underlying pattern.
So, someone more "intelligent" is akin to a camera with higher resolution. Consider why this matters, and why: According to Wolfram's self-pretentious "new kind of science", complexity goes like this:
0 : Simple, static. No pattern.
1 : Predictable, boring, oscillations. Obvious patterns.
2 : Fractals, scale invariant common features (pattern recogniseable at most scales)
3 : Patterns that seem just like natural chaos, but which spring, unexpectedly, from small Kolmogorov complexity systems.
4/inf, the real world : "natural" Chaos, no obvious pattern, maybe no true pattern/underlying system at all:
Only able to recognise a pattern if in possession of *all* the data, from the beginning... otherwise, only guesses are possible... we don't know if 4 is a case of 3, or vice versa, any more than we know if P = NP.
A "more intelligent" person can see more of the chaos, and perhaps enough to notice another class 3 pattern IRL... leading to better science, leading to more capable engineering.
So then that got me thinking about how to fix the patent system:
And as far as I can tell, the only way is to tear the whole edifice of IP law down, and build anew.
Globally.
When re-engineering systems, you really only get one chance to "do it right with version 2.0". (examples of doing it right: Mac OS X. X = 10, and what's 10 in binary?).
So... what we really need those patent office bureaucratic type people to be doing: Maintaining a centralized database of mass-produced designs, preferably in one open 3d format capable of including all details at the "as manufactured" level.
Why? Because searches for "novel features" can then be efficiently automated.
This way, design rule checks can be run (at submission) using supercomputers, or even over the entire database...
As dangerous failure modes are discovered (eg, as by air crash investigations) then those rules can be updated, and other potential disasters avoided before they happen.
The legal change would be to cancel all "patents as a license to extort" (as they are) and instead force disclosure (potentially confidential, to the registry) before production.
Then "patent style" re-numeration can be fairly based on the improvements in actual designs.
With fair limits on how that goes as well, so that small people can't be extorted out of their freedom to experiment.
Finally, in cases where centralized submission makes no sense - simply require plans to be supplied / made accessible with each one-off sale. This rule could easily apply to software: It's standard for most free-as-in-liberty software anyway.
In case this sounds too risky for the big players: The centralized database can be "confidential for a time", so that proprietary secrets are by default able to be kept... for a short while.
But this way, it becomes impossible as a matter of law to keep the recipe forever secret: At the very least, the centralised design registry must see it in order to apply those design rules.
No patents == no reason not to protect everything as trade secret, so the above would be needed to force public disclosure enough to keep everyone safe.
And the get out of fee free card: Open source the design/ or just give it to your buyer. (This is required someplaces anyway: House plans submitted to council for building design rule checking, or in electrical cabinets attaching to the grid: AS/NZS 4000 series requires electrical cabinets to contain a circuit diagram as built so that documentation is available for whoever opens the box to safely repair it.)
This way, you can sell anything, if you give the fair recipe / actual plans to who buys it, or at least into care so the buyer can be sure that someone, somewhere is checking for hazard.
And hazardous substances? Simply require MSDS disclosure with shipment. Legalize everything so long as that holds, preferably with graphical disclosure for particularly nasty things. (eg, crokodil, if you're really that desperate...)
Oh, and copyright: Should only be applicable to physical copies/knockoffs, and only to avoid loss of data. (which copyright did do for books, but it's global social benefit is gone if it's stopping people properly back them up now, when libraries are burning books they can no longer afford to keep....)
Apart from that, protection for the author is as simple as anti-fraud, which already exists. The 'net plus a timestamping, decentralized bitcoin-like logging system is enough to properly establish true priority.
Allowing "intellectual property" to be traded was a big mistake. Why should it be?
It only allows for extortion racket's like MSFT's, in the OP: The patents weren't secret... but their current "owner" was...
Patents, being "get out of jail free for specific cases of extortion (which is otherwise strictly illegal)" are the superweapon of civil law: The only defense is a good offence, or obscurity.
Except for patents, they're like nukes planted at specific spots, and you don't necessarily know how troll-like the owner is feeling.., or even who the owner is... Or even if you're near one...
(Score: 2) by SlimmPickens on Tuesday June 17 2014, @10:01AM
thanks for that ramble, I enjoyed it!
(Score: 1) by cafebabe on Wednesday June 18 2014, @03:17AM
I'm honored that you provided such an extensive reply to my comment and I feel obliged to respond; possibly in multiple messages.
From the outset, patents were a problematic covenant between state and enterprise. However, they exist and they are widely recognized. The absurdity of the situation, though, is that no-one can ever be sure that they are patent compliant. There is no service which takes a file and returns a list of patent infringements. This isn't even possible for copyright, so we're a long way from solving the problem (copyright infringement) or the meta-problem (patent infringement). Many would say neither is necessary and some cite emerging economies as examples where such regulation doesn't exist or is widely ignored. But how do we incentivize people in a balanced manner? I don't know.
Regardless, I believe the problem with patents being written in legalese is an unavoidable symptom of a mixed economy. (Without the state allowing private enterprise, there is no need to grant patent rights. Likewise for private enterprise in absence of the state.)
A personal observation of mixed economy activity across law, patents and driving examination is that junior agents are on one side and senior agents are on the other. In the case of law and driving examination, junior agents work privately and the best (as determined by their career trajectory) are recruited by the state. However, in the case of patents, I've noticed that patent examiners employed by the state tend to be young turks. Meanwhile, people filing patents tend to be wily old foxes. As an example, the most famous patent examiner, Albert Einstein, left the Swiss patent office when he was 30 but successfully filed patents after this.
I don't believe that it is a case of letting obtuse patents through because they are written in the usual doggerel. Anomalous patents are written by nutjobs while the majority of patents are filed by experienced agents who are funded by enterprises with deep pockets. If a typical patent fails, the claims can be weakened and it can be submitted again. This can be done repeatedly. So what is the consequence for a civil servant if it is allowed through first time? Indeed, when patent offices are part of an international pissing match, where is the incentive to block any patent?
Evn when we ignore a race to the bottom, a central registry in its current form doesn't work. The language in patent claims is defined in the body of each patent. Therefore, even if people use terminology in an honest manner, it is possible for patents to make an identically worded claims and cover different inventions.
My suggestions to fix the mess may be unworkable but should start with increasing the level of novelty and making applications into a one-shot proposition. I would also suggest compulsory purchase in which useful patents are purchased by the state for the good of a nation. In the case of a nation with socialized medicine, purchase paid in annual installments may increase patient privacy while also reducing cost.
1702845791×2