Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×

Comment Re: Yeah sure (Score 0) 371

by Arker (#47303445) Attached to: Court Releases DOJ Memo Justifying Drone Strike On US Citizen
" Finally, I'd also like to point out that while the US abides by the Geneva convention and other treaties and accords, it is not a signatory of the same, so while the Geneva convention sets good rules, the US is only bound to them as long as they voluntarily comply."

This is absolutely incorrect.

The US is a party to all 4 Geneva conventions, and one additional protocol (protocol 3.) These have been ratified by the Senate and are law of the land.

Comment Re:Strawman (Score 1, Insightful) 270

The benefit is only asymmetric if you (as Comcast appears to do) define 'fulfilling our contractual obligations to our customers' as a non-benefit.

Comcast sees more benefit in refusing to provide the service they are being paid for than in living up to their obligations, and that's the big problem here. The fact that they have no effective competition for most of their customers is a big part of why. The fact that they provide their own services at higher margins that compete with third party services accessible on the internet is the other big piece of it.

Both problems could be solved at once by simply making it law that ISPs have to be ISPs, and cannot be part of a larger business. Existing conglomerates like Comcast would have a period of time to spin off the ISP service which would from that point forward be ONLY an ISP and statutorily disallowed from acquiring or being acquired by other sorts of businesses.

And yes, I am a free market 'fanatic' so to some this will be a shocking view from me, but 1. the existing market is far from free and 2. a simple statutory restriction is a lot better than giving more regulatory power to the bureaucrats which will only be captured.

Comment Re:Actual Math + AP CS teacher here (Score 1) 155

by Arker (#47293551) Attached to: Computational Thinking: AP Computer Science Vs AP Statistics?
IMOP the two subjects that are critical here and get neglected are logic and number theory. Logic typically gets covered as a first or second year course in college - after many years of courses that depend on it. Teaching it much earlier would make more sense and result in better comprehension of other subjects.

Number theory is, in my experience, the branch of mathematics most relevant to programming. Far more relevant than either calculus or statistics, and (for me at least) much easier as well. Like logic it's really foundational for a lot of other stuff. Yet it's essentially unheard of. No?

Comment Re:Sorry but... (Score 0) 143

by Arker (#47290601) Attached to: Open-Source NVIDIA Driver Steps Up Its Game & Runs Much Faster
"So, again, where is anyone but you claiming things don't compile?"

It's not something you can contest, it's known and obvious. You cant compile it because you do not get the source.

The source for a shim whose only purpose is to load a binary into the kernel where a driver should have gone instead is certainly not a substitute, let alone a good substitute.

"And to be frank, I care a LOT more about things working, than being able to examine 100% of the source code."

It's a false dichotomy you (and many others) setup here, trying to claim pragmatism. It's not pragmatism though, it's simply short-sightedness combined with laziness.

"X works well enough for me, with my stuff, right now, today." There's your 'pragmatic' justification for being lazy. A real pragmatist would be concerned about how it would work with the next stuff he gets, and the stuff after that, and someone who is not intellectually lazy would certainly have some serious concerns about corrupting an otherwise Free system by integrating a black-box binary directly to the kernel, to say the least.

Frankly this is the sort of thing that should cause alarm bells to go off inside your head so hard you fall to the ground clutching your bleeding ears - solely as a pragmatist. If you had any intellectual commitment to Free Software that might intensify it even further, but it's certainly not needed.

Comment Re:Sorry but... (Score 0) 143

by Arker (#47290339) Attached to: Open-Source NVIDIA Driver Steps Up Its Game & Runs Much Faster
"Unless you're using a platform that isn't a PC"

Oh bullsquat. A PC is a Personal Computer. I am in my fourth decade of using PCs, and I have used PCs with many diverse architectures, build around chips ranging from the Z80 and x86 to Motorola 68k, MIPS, Alpha, etc. Hardware changes almost as often as skirt lengths, and often for similar reasons. A system that will work in that environment MUST include real, human-readable code, not a brittle binary written for a single monad out of the entire universe of hardware and software combinations.

What you were trying to say is 'unless you are using something other than the currently popular x86 compatible intel/amd hardware' which leaves a LOT more room outside it, but even that is not actually accurate - even if you are using exactly that it STILL wont work properly. You can claim otherwise all you want, but you admitted it does not compile so you are only contradicting yourself at this point.

That's a show stopper bug right there, absolutely unacceptable. "Works as long as you do not use the system" does not describe something that truly works. Compiling the kernel is basic, if you break that, fundamentally speaking you have broken the entire system.

Comment Re:Sorry but... (Score 0) 143

by Arker (#47289919) Attached to: Open-Source NVIDIA Driver Steps Up Its Game & Runs Much Faster
"I can't even find one supporting nouveau over nvidia's own for stability and performance."

And you are still pretending it provides that? When you just admitted it wont even compile? How do you maintain such doublethink without snapping something inside?

Seriously, out one side of your mouth you pretend to acknowledge the points I just made, but out of the other you blather on about the 'superiority' of a solution we just established flat out does not work.

And then you deride their efforts to put out a system that actually has a proper driver for your hardware, as somehow 'philosophically pure?' I am not sure why that would be a bad thing but it's ludicrously out of place anyway. Real drivers should always be preferred precisely because the end goal is to practical - ensuring we can make our boxes work, today and tomorrow.

As we just established, you absolutely cannot do that with a binary.

Comment Re:Everything is an algorithm (Score 1) 263

by Arker (#47289763) Attached to: The Supreme Court Doesn't Understand Software
"There's no mention of them in the statute, and it's solely due to SCOTUS interpretation."

SCOTUS' logic here is good. Patents are specifically for inventions, not discoveries.

" Accordingly, it may be an error to read too much into that interpretation, and suggest that no method performed by a computer could ever be patent eligible. For example, the Alice Bank opinion seems to suggest that a method performed by the computer that improves the working of the computer would be patentable - e.g. data compression methods, encryption schemes, data transmission improvements, etc.: "[Alice Bank's] method claims do not, for example, purport to improve the functioning of the computer itself.""

It's an interesting passage but I think you read too much into it. They are simply distinguishing the case they are ruling on from the cases they are not ruling on.

If a patent purported to improve the functioning of the computer itself, perhaps it would be valid. Or perhaps not. Being that the patent in front of them in this case did not, they did not rule on that question. (And reading the decision, they do show more nuance than most give them credit for. It's quite possible if such a case came up they would be able to understand that software is math by the time they ruled on it.)

Comment Re:Sorry but... (Score -1) 143

by Arker (#47289615) Attached to: Open-Source NVIDIA Driver Steps Up Its Game & Runs Much Faster
"No it really isn't. i run it on several linux boxes"

Not surprising with a small sample and homogenous hardware, and not particularly meaningful either. Others find it much less reliable.

Now try just getting it to work, forget about not crashing, but to work at all on a new kernel, or a new architecture. It aint happening.

This is what happens when you accept non-free drivers. Maybe, absolute best case, it works fine for a subset of users. It will never serve the rest, and we will never be able to fix its bugs, extend it, improve it, or maintain it. If you use it you lose the ability to upgrade or improve your own system as well.

Far too high a price to pay for fancy graphics. But fortunately you dont have to pay that price. Nouveau is more than good enough for most people right now (much like the blob) but it does not share the long-term disabilities.

Personally I have found that AMD and Nvidia usually work well enough for me using real drivers, and Intel works flawlessly. Intel has sold a few more units as a result. Good to see nouveau making progress though, it's always good to have more choices.

Comment Re:Everything is an algorithm (Score 1) 263

by Arker (#47289439) Attached to: The Supreme Court Doesn't Understand Software
The process language was aimed at chemistry originally, with the aim to make patentable novel and useful chemical processes. The courts have recognized it would be idiocy if read literally and generally declined to do so. But even with it in, it is not a free pass on the rest of the requirements, one of which is that mathematical algorithms are explicitly excluded.

So you have one part of the analysis that says maybe (a process could qualify - but not every process does) and that cannot be used to override the definite 'no's (mathematical algorithms are not patentable and a general purpose computer does absolutely nothing else.)

Comment Re:Everything is an algorithm (Score 0) 263

by Arker (#47289299) Attached to: The Supreme Court Doesn't Understand Software
"On the contrary, processes are one of the statutory categories in 35 USC 101, and algorithms are processes. The only ones that aren't patentable are the subset of "mathematical algorithms". "

We are talking about mathematical algorithms. That's the only kind a computer can process.

"You could embody the algorithm in an FPGA and it would be a patentable device... But when you have the same exact algorithm in a self-configuring processor, it's not? How does that make sense?"

Because a specific device may be patentable subject matter, while the algorithm itself is not, and the combination of an algorithm with a general purpose computer is an obvious application of the general purpose computer as soon as the algorithm is known.

Seriously, how does that not make perfect sense?

Comment Re:Sorry but... (Score 0) 143

by Arker (#47289049) Attached to: Open-Source NVIDIA Driver Steps Up Its Game & Runs Much Faster
So you happen to have one of the cards it crashes on. I can understand being annoyed by that.

But look at the big picture, please. The binary driver is a buggy POS on any system, and will never be anything else. The nouveau driver gives a viable way to support this hardware properly. Sure it still crashes on some cards. Bug reports are needed to identify the issues and fix them. The point here is this is software that CAN be fixed, unlike a binary that cant even be USED in many cases, let alone fixed.

"Add on top of that the fact that Mint devs also removed Ubuntu's boot menu option to install Linux before X starts,"

Well ok that part of it is truly retarded, I agree. Requiring X for your installer is a real rookie move and it's hilarious how many supposedly competent distros fall for it.

Comment Re:Everything is an algorithm (Score 1) 263

by Arker (#47289005) Attached to: The Supreme Court Doesn't Understand Software
"Sure, the general purpose computer is known, but if the algorithm is not, and the patent claims the computer performing that algorithm, then it's not blindingly obvious."

That's right, the algorithm itself may not be obvious, but algorithms are not patentable! That's settled law.

The patent attorneys try to get around that by patenting a "method" or a "device" instead, which just happens to embody the algorith. But when you read carefully there is no method (other than hiring a coder to program the algorithm into the computer) and no device (other than a general purpose computer doing exactly what it was designed to do.

If they would actually write an application for a special purpose machine that only implements this one algorithm (instead of any algorithm you want to hand it) and that machine itself incorporated innovations, then those should probably be patentable. But that would require actual R&D expenditures, and risks, and still would not be the gold mine that a patent on the simple idea of using a ubiquitous general purpose computer to execute a known algorithm could be.

I can understand why patent attorneys want to make more money more easily, but why should society be forced to pay the cost?

Comment Re:Program == Theorem (Score 1) 263

by Arker (#47288941) Attached to: The Supreme Court Doesn't Understand Software
"When you see some output on the screen, do you know which part of that output is generated by hardware and which part by software? Obviously not."

Obvious nonsense. It's all 'generated' by the hardware. By a general purpose computer doing what a general purpose computer does - processing input and producing output according to fixed mathematical rules.

"Anything you can do in hardware, you can do in software and vice versa."

Not true. The one and only thing you can do 'in software' is read input perform mathematical operations and output the result. Any effect that you want to produce in the outside world (aside from trivial side-effects) has to be done by producing the correct number and sending it to the correct output hardware, which actually produces the effect. Not the computer itself.

Now it's true, you can produce the number using a program on a general purpose computer, or you could instead construct a special purpose machine that does only this one algorithm and does it much faster. So what? That has NOTHING to do with patent eligibility, or with whether or not the algorithm is math. A special purpose machine might be patent eligible, as it might incorporate innovations that for instance bring the cost of implementation down considerably and let them compete where they previously could not. Fine. That does NOT mean that the algorithm itself is patentable, or that using a general purpose machine to execute the same algorithm is patentable (which in the modern world would be extremely close to, perhaps indistinguishable from, the same thing.)

Comment Re:Program == Theorem (Score 0) 263

by Arker (#47287857) Attached to: The Supreme Court Doesn't Understand Software
"There are plenty of patented electronic (hardware) and mechanical machines that do what humans can do by hand. That doesn't make them unpatentable."

No one said computers are not patentable. Let me repeat that. No one said COMPUTERS were not patentable. We are talking about software. And software is not the computer. Software is a number to be given the computer as input. Very different things.

Comment Re:Followed the law. if (false) then false (Score 1) 263

by Arker (#47287093) Attached to: The Supreme Court Doesn't Understand Software
"It's been argued that had Huffman implemented his famous compression algorithm in hardware rather than software he would have been granted a patent."

Which would be (relatively) fine. Because if he implemented it in hardware, and got a patent on that, it would only cover the specific techniques he came up with to facilitate implementing it in hardware. It would not necessarily cover all hardware implementations (it would almost certainly be possible to design another hardware implementation of the same algorithm by doing it differently in at least one key respect) and it would certainly not cover using a general purpose computer to implement the algorithm (a completely obvious step once the algorithm is known since the one and only purpose of a computer is to implement algorithms.)

Slashdot Top Deals

How many NASA managers does it take to screw in a lightbulb? "That's a known problem... don't worry about it."

Close