Microsoft chucks patent sueball at Kyocera over Android phones Microsoft has filed for an injunction to ban Japanese mobe-maker Kyocera from selling its Duraforce, Hydro and Brigadier phones in the US, alleging a breach of its Android patents. The Redmond-based software giant asked the US federal court in the Western District of Washington to impose an injunction against Kyocera, based on …
Well, they have shown one key patent now. US7289102 tries to patent seqlock
Seqlock was originally invented as a part of ntp and has been in the Linux kernel source tree since from around 2.5 - 2.6 which dates it ~ 2001 or thereabouts.
The priority date on the patent is 5 Apr 1999, so on the surface it is valid. Now the fact that US patent system allows to file in 2004 for something backdated 5 years is frankly outrageous, but that is different story.
Personally, I have worked extensively with this after 2010. Instantly recognized what it is and what are the "civilian applications". However, I do not know the original history of the sequence locking in Linux and BSD so someone else needs to come out and say where did this come from.
The more interesting part is that Microsoft has _FAILED_ to pay the support fees (status shows as lapsed) for this one, so actually, it cannot ask for money on it in court. The court filing is thus invalid, end of story.
You have to be able to provide some sort of proof for the priority date on a patent. For software patents, I'd assume you'd need to have your source control system showing checkins at a certain date.
Ignoring this particular patent and software patents in general, do you really think Microsoft would be trolling Linux kernel source and backdating patents based on what it is added? No matter how evil you may think they are, they're a business first and knowingly and openly lying about priority dates in a patent would get them in huge trouble. Their lawyers will require the engineer provide such proof before filing it with the USPTO.
do you really think Microsoft would be trolling Linux kernel source and backdating patents based on what it is added
Not likely. More likely someone much later during the work on virtualization came across the timekeeping code (you have to deal with it when you work on virtual like it or not) and said - hey, we have something like that.
Otherwise, seqlock is an interesting approach. Its usefulness in real life is limited to gettimofday as it requires reader to writer ratios in the hundreds (if not thousands) to be of any use. Obtaining the time in an SMP system is the sole application that allways has this access pattern and the only application where you can safely use it and have a performance gain.
In any case Microsoft has _FAILED_ to pay for the maintenance of this patent, they are not _ENTITLED_ (at least by EPO rules, and probably by USPTO rules too) to bring it up in court. They can bugger off on this one - end of story.
Micro Soddit for Mobiles (tm) - a development philosophy used throughout the business, tried and tested for generations. A simple to follow approach which clearly gives huge profits as a quick look at the balance sheet shows.
Anyone can adopt it - you don't need actual patents, just a long history of FUD and an investment of the profits from previous extortions into more legal coffers.
In a nutshell: Shall we innovate or litigate? Soddit, we get more by using lawyers than developers so just stick to the usual.
We probably need to take a fresh look at patent terms in the modern age. Technology moves so quickly now that what was once a suitable time period to recoup one's investments and rewards can actually cross a whole generation of technology and into the next.
The Java patent in the above article was filed in 1997. About eighteen years ago. If you were the person who invented the lightbulb (I don't know who that is, I just suspect it wasn't Edison), then eighteen years is quite frankly probably too short a time to be properly rewarded for such a novel and useful creation. But whilst the Java patent was probably a clever little programming solution back in 1997 and maybe even novel and patent-worthy, we have seen an entire generation of technology rise and fall in that timespan and today such a patent seems merely to hold things back. It overlaps with too many other parallel approaches invented since then.
You can't make them too short - it can take years to bring a product to market: Hololens has things in it that are patented I'm sure and it's been in development for a long time. It would not be fair for all that research to expire just as the product was coming to market so that others could simply rip off all the hardwork and copy it for rival products that had no R&D costs to recoup. But equally you can't make them too long. And there will never be a one size fits all perfect value.
Probably we need a more sophisticated approach that considers what is actually being patented and calculates a period based on time to market, level of research required to produce and the level of innovation.
Each one of these things is dangerous because it starts to bring in subjectivity. What is the "level of innovation" on a scale of 1-5? Some things are likely to be invented over and over again. The Java patent might well be clever. But when it's "steam engine time" (i.e. the conditions are right) someone else would have come up with it in the intervening 18 years in one form or another. Whereas something else such as a new encryption algorithm or some of the things going on in materials science are things that would have a very good chance of nobody else coming up with them in the same time period, because they're very distinct ideas.
The first thing to do with patent timescales is to acknowledge that whatever you set is going to be an approximation. Only then can you start to work out a more nuanced approach. Whatever we come up with is not going to be fair, that we know. But this does not stop it being an improvement on an absolute and arbitrary value.
its also a case of COST. you can come up with a nifty little thing like the Java case over a coffee chat in the morning and have PoC and working code by the end of the day. what are you actually expecting for *that* ROI ? should such trivialities even be patented so that you suck the life out of innovation - as others waste time trying to find a trickier way of doing things or make their product less useful to avoid using that method and thus don't deliver other things? these ideas are DAMAGING the ecosystem, not aiding it or humanity.
Well you can't directly tie it into cost of the research in any proportional way. Suppose I spend a few years with my team and several million pounds and come up with a slightly different take on asprin. It's not really any more effective but it's a new drug so it's patentable. If you factor the lifespan on cost, it would last a long time. Whereas suppose in a flash of brilliance one weekend I suddenly come up with a new encryption algorithm. Something brand new and five times harder to brute-force than AES. Based on cost, that would have very little lifespan but is (probably) vastly more valuable and not only that, it might be something that no-one else would have thought of.
That's why it needs to be determined by multiple factors. And on the basis of most important factor, rather than all additive. So in pseudo code something like this:
Lifespan = K * MAX(researchRequired, innovativenessRequired).
If it is something that would take a lot of research / testing to create, that counts. If it is a unique insight unlikely to arise as a natural extension, that can count instead. Both are extremely problematic of course. Note that I wrote Research Required, not Research Done. You don't want to reward inefficiencies nor allow people to pad things out by including irrelevancies. Similarly, many things that are genuinely innovative look obvious in retrospect.
Fundamentally, we have one great glaring problem in creating any working and fair patent system - you need to be an expert to judge a patent's merit and there are few to none independent experts.
Maybe we can re-purpose academia as patent consultants? Most academics are lousy at actually getting anything done but they're great at critiquing and understanding stuff. Perhaps that is the resource we should be exploiting for this.
That's a problem with allowing software patents at all. Software patents that solve a particular meatspace problem, maybe. Software patents that are merely a technique for doing something with a particular piece of software (i.e. Java classes) are ridiculous and should never have been allowed.
"Combining multiple java class files into a run-time image” - How can you patent something that has to have "prior art" ? I mean, to run Java programs, you need a JVM ... and the JVM itself is very first implementation of a "Combination of multiple java class files into a run-time image” ... next they'll claim "We created Java" ... Now, how can anyone grant that patent ?
The other patents are just as silly - somebody must sue the hell out of them and the patent system for allowing this non-sense.
MS want to take one of a very small number of phone makers thinking of actually making Winphone mobiles? Taking them to court over Android phones. Is that a really pants way of making them do windows phone only?
Maybe the agreement they reach is that whatever rugged phone they make, they will also do a windows phone. I hate to think of this as a M$ tactic to simply get someone else other than Nokia to make windows phone.
More importantly, M$ own patents in Android so surely it is in their interest to do stuff in Android. So why are they so resistant and have been for so long. Only now have the awful 365 office versions started to come out in the play store, where is the rest of it? (I know they want people to buy the Winphone, but as the numbers are tiny, why not embrace android as well?
Patent law is pants.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
