Before anybody gets too excited, know that this case doesn't actually open the door to invalidating software patents in general. It clarifies things a little bit is all.
The points he made about how to improve the patent system are really good ones. I have always though a short lifespan on patents would be better for inovation.
I think patents are good for certain things but over all it slows inovation. I think we could be another 25 years farther down the road with technology if there where not so many bad patents out there.
The problem with merely shortening the lifespan of patents is hinted at by the logic in the argument for doing it. When we grant a monopoly to a party in return for a patent, what we are in essence saying is "If you hadn't filed here then we would not know of this invention until later."
This argument made sense when the system was first conceived because the norm for that time was to keep inventions as family secrets and innovators themselves were rare. But these days the norm is for some openness and there are so many innovators that there can be very little margin between first to invent and second to invent. Software patents in particular are problematic because it is far more open by its very nature -- the existence of a solution highlights the search space. The problem with merely shortening patents in this light is that it just adjusts the cost/benefit calculation on the basis of this being a sudden jump when this development is the result of a trend that has no signs of stopping.
Fortunately, the revision seems straightforward: revamp patents to make the calculation explicit. By the time software patents are completely useless, they will have a 0 year life (and be dead). Meanwhile if other fields end up as amendable to optimization as software is, we can gracefully scale down the patents there without having the same conversation all over again.
The best thing for everyone at this point is for individuals (and especially for small start-ups) not to file for patents at all.
This is not the best thing to do for the individual, especially if he wants to be acquired. The best thing to do in that case is to file as many broad and unspecific patents as possible to boost your perceived value. Good for you if you get acquired (or go bust and fire-sale everything), bad for everyone else who now has to live with even more bogus patents that are likely on their way to troll-land.
Its a case of "see, this is why we can't have nice things". It is conceivable for the patent system to work if there weren't so many bad actors maximizing their own gain at the multiplied expense of everyone else. Even those with good intentions feel that they have to do this a little just to get by.
"That said, it's the world we live in. So, like every other Silicon Valley entrepreneur, I file patents." Everyone feels that they have to graze their sheep to survive, but its making the common grounds into desert.
I don't know if this has been mentioned previously, but I've not seen it in any of the previous debate about software patents.
Quoting from the article:
Patents are meant to foster innovation by protecting the inventor and then upon expiry providing a library of information for others to build on. To this end, going forward, if a patent doesn’t actually tell you enough information to understand and build the invention, it shouldn’t be valid.
Given that a software patent is...well, software, why not require that actual working code be submitted as part of the patent application/grant? The code is held in escrow/confidence by the patent office for the term of the patent, and is released to the public domain when it expires.
There's probably some issue there with copyright of the code that might complicate the public domain part, but at the very least you could clean-room reimplement it. An alternative might be that you are required to renounce/void copyright on the code in order to obtain a patent - the short term monopoly of the basic idea comes at the cost of no long/indefinite-term copyright on the specific implementation.
That opens up all sorts of new possibilities, like requiring/permitting the patent claims to be machine-readable, in the sense of a conformance test-suite. That both ensures that the provided implementation is a true instance of the patented invention, and gives the holder/patent office/court an easy way to check for violation.
I really like that idea, I am sure I am missing some important part that makes it not so easy to do. When you file a patent for invention like (dishwasher) you have to supply all the design drawings for the project. I do not see why they could not do that with software patents.
Notice that the drawings for that showshoe binding patent are not "design drawings" for a snowshoe binding. (Those drawings are lacking a lot of dimensions that one would need for such a product.)
You have to provide information that tells someone skilled in the art how to do what you claimed. That's it.
FWIW, "What you claimed" is rarely a complete product. Instead, it's an aspect that can be used in multiple products. (While I don't know where else one might use that snowboard binding mechanism, I'm sure that someone else does.)
"Actual, working code" might be prohibitively context-dependent. Working code to detect credit-card fraud would be pretty dependent on your financial infrastructure. In many cases though, that would not be as big a problem, and in others it might be satisfactorily overcome. I think this is a good idea.
Yeah, it did occur to me after commenting that it might be difficult to extract the specific patentable invention chunk from the overall application.
I can't think of a single approach that would work everywhere, but logically isolating it as a library or service might work, similar to the way in which you can implement an IPC/RPC shim for interfacing to GPL'd code from your proprietary app.
Inventions that don't fit that model (Operating Systems, Embedded Firmware, Massive Scale, etc) complicate the matter, but I imagine a set of disclosure options could be devised, and the inventor/patent office would select the most appropriate.
Also what about code in a non-standard language? Suppose no one knew about Brainfuck and you included your software solution implemented in Brainfuck without an explanation of the code and how it corresponds to the patent. Do you need to include compiler/interpreter details, explain what the machine code generally looks like and does? What about hypothetical computing devices like quantum computers?
Though a general solution to proprietary hardware that you have rights over would be to include an FPGA implementation... whether that should be required or not is another story, especially if you don't have rights over it.
patents are forever... don't forget what the gov't considers the most important concept of patents; Taxe$. Patents were designed so the gov't could tax any proceeds gained by selling or licensing the patent. Why would they ever voluntarily get rid of that? They would patent their own mother if they could, or haven't they done/tried that already? http://www.nytimes.com/2010/03/30/business/30gene.html
The whole idea of "fostering innovation" is all BS. It's all about the $$$.
16 comments
[ 409 ms ] story [ 439 ms ] threadI think patents are good for certain things but over all it slows inovation. I think we could be another 25 years farther down the road with technology if there where not so many bad patents out there.
This argument made sense when the system was first conceived because the norm for that time was to keep inventions as family secrets and innovators themselves were rare. But these days the norm is for some openness and there are so many innovators that there can be very little margin between first to invent and second to invent. Software patents in particular are problematic because it is far more open by its very nature -- the existence of a solution highlights the search space. The problem with merely shortening patents in this light is that it just adjusts the cost/benefit calculation on the basis of this being a sudden jump when this development is the result of a trend that has no signs of stopping.
Fortunately, the revision seems straightforward: revamp patents to make the calculation explicit. By the time software patents are completely useless, they will have a 0 year life (and be dead). Meanwhile if other fields end up as amendable to optimization as software is, we can gracefully scale down the patents there without having the same conversation all over again.
This is patents-as-positional-good, a zero sum game that consumes resources and produces effectively no net value.
(Of course patents are problematic, but the reasons that I know of are not related to the tragedy of the commons in a way that I'm aware of.)
This is not the best thing to do for the individual, especially if he wants to be acquired. The best thing to do in that case is to file as many broad and unspecific patents as possible to boost your perceived value. Good for you if you get acquired (or go bust and fire-sale everything), bad for everyone else who now has to live with even more bogus patents that are likely on their way to troll-land.
Its a case of "see, this is why we can't have nice things". It is conceivable for the patent system to work if there weren't so many bad actors maximizing their own gain at the multiplied expense of everyone else. Even those with good intentions feel that they have to do this a little just to get by.
"That said, it's the world we live in. So, like every other Silicon Valley entrepreneur, I file patents." Everyone feels that they have to graze their sheep to survive, but its making the common grounds into desert.
Quoting from the article:
Patents are meant to foster innovation by protecting the inventor and then upon expiry providing a library of information for others to build on. To this end, going forward, if a patent doesn’t actually tell you enough information to understand and build the invention, it shouldn’t be valid.
Given that a software patent is...well, software, why not require that actual working code be submitted as part of the patent application/grant? The code is held in escrow/confidence by the patent office for the term of the patent, and is released to the public domain when it expires.
There's probably some issue there with copyright of the code that might complicate the public domain part, but at the very least you could clean-room reimplement it. An alternative might be that you are required to renounce/void copyright on the code in order to obtain a patent - the short term monopoly of the basic idea comes at the cost of no long/indefinite-term copyright on the specific implementation.
That opens up all sorts of new possibilities, like requiring/permitting the patent claims to be machine-readable, in the sense of a conformance test-suite. That both ensures that the provided implementation is a true instance of the patented invention, and gives the holder/patent office/court an easy way to check for violation.
Actually, you don't. Look at some patents, such as http://www.google.com/patents?id=TdUkAAAAEBAJ&zoom=4&... .
Notice that the drawings for that showshoe binding patent are not "design drawings" for a snowshoe binding. (Those drawings are lacking a lot of dimensions that one would need for such a product.)
You have to provide information that tells someone skilled in the art how to do what you claimed. That's it.
FWIW, "What you claimed" is rarely a complete product. Instead, it's an aspect that can be used in multiple products. (While I don't know where else one might use that snowboard binding mechanism, I'm sure that someone else does.)
I can't think of a single approach that would work everywhere, but logically isolating it as a library or service might work, similar to the way in which you can implement an IPC/RPC shim for interfacing to GPL'd code from your proprietary app.
Inventions that don't fit that model (Operating Systems, Embedded Firmware, Massive Scale, etc) complicate the matter, but I imagine a set of disclosure options could be devised, and the inventor/patent office would select the most appropriate.
Though a general solution to proprietary hardware that you have rights over would be to include an FPGA implementation... whether that should be required or not is another story, especially if you don't have rights over it.