Slashdot Mirror
Tim Bray Finds An Affinity Between Patents And OSS
Manuzhai writes "Tim Bray, of XML co-invention fame, is writing about software patents and Open Source software today. While he deprecates the 'business-method' patents like one-click ordering, he thinks some (Open) source code could tell the truth about a patent application: 'In fact, in an ideal world, I'd rewrite the law to allow software patents but require a working Open-Source implementation as a condition of getting one.'"
Well, with Novell now throwing its substantial patent portfolio behind open source, and Microsoft having promised they won't use their patents to crush open source, I don't see what all the fuss is about.
Perhaps if slashdotters relied on a software patent as their primary source of income, as I do, they wouldn't be so critical of them.
I think a way to handle the patent problem
is to make a patent license which work with patent law
as the GPL work with copyright law.
It's a way of pinning down patents to a specific algorithm. Some (most) of these applications are so broad it's difficult to figure out what the patent covers and what it doesnt. A working implementation makes this 100% clear.
If you use the code, if you use their patented stuff, you had best negotiate a license or be sued out of existence. However, if you want to code around the patent, this could be very useful.
HBI's Law: Frequency of calling others Nazis is directly correlated with the likelihood of the accuser being Communist.
Just how long do patents last on software, 75 years(ish)? It strikes me that innovation should be rewarded by a short patent of say 7 years then the IT community benefit straight after.
Besides, this proposal has a problem of its own: software patents can include so-called "program claims". In that case, not just the use of a program which infringes the patent, but also the publication of such a program can be forbidden. Many software patents (and even non-software patents) are starting to include those.
If you start adding source code to patents, then those patent descriptions themselves can infringe on other patents. It's completely silly since the patent system is supposed to encourage publication of information, but nowadays it can also be used to for forbidding publication of information (because some people think software is the same as a "machine", instead of simply a description of something).
Donate free food here
As I recall, you are also free to use patented stuff in your own home without paying for it - only commercial use would actually infringe.
IANAL, and while this seems like an attempt to weasle around the system I rather like it.
Suppose you're a keen young programmer and you've figured out a keen new algorithm for securing a communications channel or crash-proofing a database or animating an MMPORG monster.
He goes on to suggest "well, why not" a patent.
He expects us to believe that he doesn't realize that "a keen young programmer" hasn't got the tens of K dollars to get a patent, and certainly hasn't got the millions of dollars needed to defend a patent against wilful infringers.
I think the article is probably just astroturf; after all, Bray is now a Sun employee and the Sun's line is that software patents are a Good Thing.
Anyway, not all lawyers seem to be happy with this, but as Karl-Friedrich Lenz writes:
Donate free food here
And just out of curiosity, what happens when you write a piece of software which uses 11 patents, all of which are guaranteed 10% of the profits?
Bray says that software patents do a lousy job of disclosing inventions: they are (sez he) ``notoriously inaccurate, incomplete, and unreadable.'' He assumes that the patent office, which can't even seem to read the patent applications, would be able to test the software and determine whether it was indeed a functional and useful implementaion. That sounds optimistic.
Furthermore, any patented methods and the code which embodied them would be of no value to Libre software until the patent had run out. After all, public domain implementation or not, you still have the problem of a license for the method!
See what I've been reading.
Software patents, aren't in themselves bad. The thing that isn't addressed is their timespan.
Patents for physical things took into account the fact that they needed to be drawn, engineered, factories built to create them, distribute them, have them installed where necessary, and then cover them through a fair lifespan.
Take, for example, valves on a chemical plant. A new and innovative one could be thought of and patented.
Then, the factories set up to produce it (say a year from patent perhaps, now safe to give the designs under contract, as it's patent protected), then it needs to be marketed, so, perhaps 2 years from inception to starting to get used. Initial tests and usage in industry, say, 4-5 years until it really starts to be used industry wide.
Lifetime of a valve, perhaps 10 years if they're in a harsh environment, more if not. So, you get in one round of replacement of the same thing.
But, the timescale there for a physical item that's supposed to last 20, 30 or more years isn't terrible. It's still VERY useful in 30-40 years.
Now, software, protected for the same duration.
Patent is drawn up. Software out the door days later, as there are no real tooling and production costs (relatively speaking). It's possible for sales to ramp up and reach market saturation within a year, if it's something innovative and useful. Industry acceptance and having it treated as 'old and established' within 2.
Within about 5 years, it's (usually) classified as obsolete.
So, for the next 20 years after being obsolete, it's holding back the market from developing it's successor, because it's patent encumbered, and license fees need be paid on it. So, the next generation flounders.
If Patents took into account the average obsolescence period of the market, and allowed a patent for the given period, things would work nicely.
Once it's in the 'getting a little old and clunky' period, anyone can then make a free implementation of it, or perhaps design it's successor based on the original. OR a proprietary new version, or whatever! But it keeps things moving, which is what patents were meant to do all along.
5 years for a software patent? Sure, that sounds fine. Maybe 7 or 8 at a push. If you've not made money off an idea in that time with a captive market in the tech game, you're probably not going to.
But the 5 years is enough to allow something to prosper, while ensuring that you keep thinking of the next idea, or allowing someone else to.
And perhaps it would stop all these patent shops churning out nothing in the knowledge that they've got 20 odd years to sit on it and hope someone comes up with something they can shoehorn into what they've got on their papers.
5 years is a lot less time, enough, really, to say "If you're not going to use it, then you've had your chance to, now let someone else actually do something good with the idea"..
In any case, the traditional model of a patent is absolutely inapplicable to software. The original purpose of a patent was a compromise deal between an inventor and a government {acting as a representative of Society}. The problem was that an inventor might have exhausted their capital developing a working prototype; if they now show this directly to an established manufacturer, the manufacturer might decide to cut the original inventor out of the loop; whereas if they approach a {non-technically-minded} banker hoping to obtain a loan on the strength of the invention, with which to set up their own factory to make it, the banker might not be convinced of its viability. So the Patent Office was set up, to give inventors an official letter precisely describing their invention, and granting them a time-limited monopoly over it. This can be used as proof that the invention works {to obtain financial backing} and as a description of how to make it {to obtain manufacturing services}, but it also specifies a date after which the invention is to be given to society at large for the benefit of everyone. If the invention is a good one, then any loan should have been paid off long before such time.
In the case of software, where the cost of reproduction is essentially nil, the inventor is not prevented by financial constraints from further developing their idea beyond the prototype. So the original need for a patent is absent.
Je fume. Tu fumes. Nous fûmes!