Slashdot Mirror
Stallman Critical of OSDL Patent Project
PatPending writes to mention a News.com article about Richard Stallman's objections to the OSDL patent project. He argues that the project may actually be 'worse than nothing', as it will undermine certain legal tactics. From the article: "'Thus, our main chance of invalidating a patent in court is to find prior art that the Patent Office has not studied,' Stallman wrote. Second, patent applicants could use the prior art uncovered by the OSDL to write patent claims that simply avoid the technologies used in the tagged software. 'The Patent Office is eager to help patent applicants do this,' Stallman wrote. Finally, he wrote, a 'laborious half measure' such as the Open Source as Prior Art project could divert attention from the real problem: that software is patentable in the first place."
I'd bet on RMS, smelly hippy though he is, being right in the mid to long term, if for no other reason than he hasn't (to the best of my knowledge) been wrong yet. In any prediction. Ever.
In purely practical terms, the OSDL patent project is like trying to put out a burning forest by standing close enough to sweat on it.
If you were blocking sigs, you wouldn't have to read this.
I actually agree with rms for the most part, but will play devil's advocate for a bit here. Stallman never liked conventional software licenses. He wanted to create and use free software but licenses got in the way. He could have fought all licenses and even all copyrights, and demanded that all information be free. Instead he built a license upon established copyright law, and the GNU GPL was born. Now he has a problem with software patents. Instead of supporting the free and open use of patents, he is saying that all software patents are unjust. Why does Stallman consider the OSDL patent initiative so bad? If it is unfair because it uses the same legal protections as the corporate trolls, then doesn't the GPL legitimate the unjust system of software licenses in the same way?
------ Take away the right to say fuck and you take away the right to say fuck the government.
-1 Analogy
No it's not like that at all. It's not like anything, except the plain statement that patents, designed for finite, physical objects, should not be applied to infinitely reproducible items like software.
Software is not a house, a car or any other physical thing. Please stop pretending it is.
Crumb's Corollary: Never bring a knife to a bun fight.
Keep in mind that the we're talking specifically of software patents here.
Drug and hardware patents are also problematic, but the reform had better be well considered, or the cure could be worse than the disease. The specific case of software, however, is one where we can eschew playing without destabilizing the economy too readily.
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
Probably the best defense against having to deal with software patents is to keep the software closed. Don't make the code public and don't tell how it works. If people don't know you've violated their patent, they are not likely to sue you, and their software patent won't be worth very much.
I hate to ask this, but if someone uses your code and uses it in their own... Isn't that a copyright violation?
I say this because just because you have the code doesn't make the process easy to copy. Unless of course you copy the code... Ergo... Copyright violation of the GPL.
No one can simply look at 50,000 lines of code and go... "OOOOH! I wish I had thought of that process!"
(well most of us anyways)
But chances are the programmers are going to by really tempted to use a copy and paste with their text utility before they can analyze the process and copy the end results.
That said, if you can simply copy the process by looking at the end results (and or code), chances are it wasn't worth patenting in the first place.
People shouldn't be paid for their ideas, but rather the implementation of them.
If I come up with an idea and use that to create a physical working device then I should patent the device... Not the idea of the device.
If I come with an idea and I write code then the code and the effort is copyrighted.
If I come up with an idea for a business... I should create the business and trademark it and not patent the idea to keep people from competing with my so called "idea".
The reasons behind patents and copyrights was to benefit society and give a carrot to the author or inventor.
The fact the author or inventor gets paid is a nice side effect of this part of our constitution, but it is not a god given right.
On occasion during the early history of our nation... Some patents were revoked for the beenfit of our society like the first non-hand wheat thresher because the PTO thought it too important.
"I am the king of the Romans, and am superior to rules of grammar!"
-Sigismund, Holy Roman Emperor (1368-1437)
I suppose that if you implement computer software which is sufficiently original, not completely obvious after 5 minutes of thought, and is not representable as a mathematical algorithm, that might deserve the protection of a patent
So was everyone asleep in that part of computational theory where they point out that everything (including hardware) is reducible to a mathematical algorithm?
Indeed. Look at crypto export regulations. A lot of people now think that cryptography isn't meaningfully regulated, but in reality, only mass-market software is free from the onerous restrictions. If you start selling, say, a secure wireless keyboard over the Internet, you can still be charged with a crime. (Why do you think there are *still* no actually-secure wireless keyboards on the market?)
http://outcampaign.org/
To compute the result, you must press a few hundred thousand calculator buttons, all of them correctly and in the right order. Would having a calculator where instead you could push just a single button to give you the answer reliably have any value to you?
Your argument is effectively saying that it wouldn't. After all, it's just maths, right? Anyone can do it and we know it all already, so there's no no need to incentivise new research.
What? You think I'm arguing that neither math nor its reliable and speedy computation are valuable? Where did you get that?
What I'm saying is that codifying mathematics into a computer program does not change the fact that it is mathematics.
In your example there, you already have the algorithm (i.e. the math), and you want to execute it quickly and reliably at the push of a button, instead of entering it in each time. Of course that has value, but that additional value beyond the math itself is 1) the specification of the math in a computer-readable form and 2) the computer to run the specification.
The first is protected by copyright, the second by patents. The math itself, which is what patenting the part of software that isn't the specific fixed representation involves, is unpatentable, and for good reason.
No need to incentivize new research? The incentive is the usefulness of the mathematics, same as it has always been. Was there no mathematics research before the creation of software patents? What about after? That's rhetorical, of course.
By the way, as another poster pointed out, your choice of Newton as an example is rather ironic. Try reading a little of the history of Newton and Leibniz, particularly the extensive delay in Newton publishing his work on calculus, the allegations of plagiarism, and the divide in the mathematical community that resulted.
I knew about the fight, but not about the 30 year delay. Learn something new every day. I just hope you realize that if the early practitioners of computer science in the 50s and 60s had locked away their creations for 30 years, we wouldn't be having this conversation today.
Math builds upon math. Science builds upon math. Engineering builds upon math. Math is the fundamental language of technological progress, and locking math up in patents is to lock up the foundation of progress.
The enemies of Democracy are
In theory, perhaps. More accurately, ideal hardware can be represented or modelled by mathematical algorithms, but real-world hardware exhibits a number of differences from ideal models, including non-perfect TTL response: real transistors aren't ideal binary on-off devices and exhibit non-linear behavior especially as they heat up or run outside of rated tolerances, not that anyone would overclock a circuit or have issues with their CPUs overheating.
Of course, you can model some failure modes, at least once you have enough data to predict the likely ones, but computer manufacturers are in the process of recalling millions of batteries outside of their predicted MTBF, just as some hard drive vendors have had to recall lots of devices because of shoddy components, and there's a whole era of AMD motherboards from the KT266 VIA chipsets through the KT400 or 600 which had those crappy electrolytic capacitors which used to fail after a year or so.
As for computer software, self-contained sections of code working over known data exhibit highly predictable results which can be predicted or modelled deterministicly, but it's easy enough to model algorithms such as the Lorentz attactor or Lyapunov exponents which may be deterministic but are sufficiently non-linear or even chaotic that their states cannot be predicted short of actually computing the results. More to the point, most software nowadays involves interactions with humans which provide more-or-less arbitrary inputs and sometimes you even gets software bugs resulting from un-expected combinations.
"The human race's favorite method for being in control of the facts is to ignore them." -Celia Green