|
Kalman posted:I'd take issue with that paragraph as well. Programmers certainly seem to come up with plenty of different ways to do things like sorting, searching, cryptography, creating pseudorandom numbers, and all sorts of other things. There doesn't have to be a One True Way, there just has to be an obvious way. Let's say you want to synchronize a bunch of programs simultaneously, how would you do that? Maybe you keep a list of all the programs you want synchronized and then synchronize each one in the list. The inventors on that patent didn't solve any hard problem, they just identified something people might one day want to do and patented the perfectly obvious way of doing it. I'm not anti-software patent per se. I have no problem with RSA. for example. That was a hard problem with a non-obvious solution. What I have a problem with is if they had claimed "an encrypted communication system that consists of a first terminal that receives a token from a second terminal which receives a token from the first terminal allowing it to decrypt the message." The standard should be, if you asked some random programmer how to solve problem X, with no knowledge of how the patent applicant solved problem X, would they come up with the same answer? If so, the answer is obvious.
|
# ? Dec 10, 2013 21:48 |
|
|
# ? May 28, 2024 14:29 |
Kalman posted:Rounded corners on a phone aren't being patented as a new useful feature; they're being patented as a new ornamental feature. Usefulness isn't a requirement for design patents - they don't fall under 101 requirements (new and useful) but instead 171 requirements (new, original, and ornamental). While I'm no expert in design patent law, I was a bit surprised that patent wasn't attacked on the grounds that a majority of the "ornamental" features were actually functional in nature. Rounded corners easier to manufacture and pull out of your pocket, a bezel all around a screen allowing for adhesion of the LCD and connection of touch input circuitry as well as permitting a user to grip the device without covering up the screen, a centered button at the bottom to allow a left or right handed user to press it with equal ease without covering the screen, etc. I think that certain utilitarian and minimalist industrial design trends raise the really interesting question of if a lack of ornamentation can itself be ornamental.
|
|
# ? Dec 10, 2013 22:12 |
|
KernelSlanders posted:There doesn't have to be a One True Way, there just has to be an obvious way. Let's say you want to synchronize a bunch of programs simultaneously, how would you do that? Maybe you keep a list of all the programs you want synchronized and then synchronize each one in the list. The inventors on that patent didn't solve any hard problem, they just identified something people might one day want to do and patented the perfectly obvious way of doing it. I'm not anti-software patent per se. I have no problem with RSA. for example. That was a hard problem with a non-obvious solution. What I have a problem with is if they had claimed "an encrypted communication system that consists of a first terminal that receives a token from a second terminal which receives a token from the first terminal allowing it to decrypt the message." The standard should be, if you asked some random programmer how to solve problem X, with no knowledge of how the patent applicant solved problem X, would they come up with the same answer? If so, the answer is obvious. But that assumes the problem was already known and just hadn't been solved because no one had gotten around to it. Sometimes that's true, but your standard doesn't differentiate between obvious solutions to obvious problems and obvious solutions to non-obvious problems. If we assume that we also want to incentivize people whose invention consists of seeing the problem no one else had thought of before but that was easy to solve once thought of. A couple of examples from the PTO include the use of a sub coating to reduce degradation of omeprazole pills when no one realized that the uncoated version degraded, or reversing the sweep of a turbofan blade (easy for an ordinary turbofan designer to do) when there was no recognition that reversing sweep would improve a particular performance characteristic. There's also a huge issue in defining X for your standard. For example, if the "problem to be solved" is "create a public key crypto system" without the RSA patent, that seems hard. If the problem to be solved is "create a system where prime numbers define a key that can be provided to the public in order to encrypt something and define a second key that can be kept private for decryption", well, competent programmer might figure it out. Similarly, for the other patent, if the problem to be solved is "synchronize a group of different programs between two computers", you probably wind up with at least a couple of similar results to the patents management program which calls libraries corresponding to individual known programs to sync the data associated with those known programs (although some people will do it via offline file sync, some might do it by integrating libraries from the programs, some might do it by mirroring file writes, etc.). On the other hand, if the problem is "synchronize two computers" then you'll wind up with a bunch of different solutions, some of which might look like the patent, some of which won't. (That particular patent does look bad; as always, there are probably limiting factors buried in the spec and file history, but it's written pretty broadly. Again, it doesn't actually cover all methods of synchronizing two computers, but it falls into what a lot of my coworkers and I call the "do it again" trap, where you get a patent on something that everyone knows how to do with one thing and you just patent doing it for multiple things or switching between multiple things.)
|
# ? Dec 10, 2013 22:19 |
|
hobbesmaster posted:Except for you know the thing in that post you quoted. The point even of things like SBIR is not to turn a profit(at least the CBO didn't say it makes money), it's to "spur innovation" or actually see some of these things get used("commercialized"). It's all kind of nebulous, and most of what I've read on SBIR has been equally nebulous* or anecdotal. My assertion is that absent the fiscal impact(which is incredibly difficult to distill) on the university you could make research public domain and really nothing changes from an encourage innovation/product view since any incremental value of these projects in that sense must be in the people, later research, and trade secrets developed AFTER the split from the University(because you already have what was developed at the university with government funding, absent any IP laws at all). * - There are a bunch of dueling papers on the impact and positive/negative nature of Bayh-Dole in general on university research if you have jstor access. Good/bad seems to break down by originating discipline and length of academic career though. Vladimir Putin posted:Everything in this post is wrong. Your concept of what happens in government sponsored research is totally off base. Just to make sure I have this right, the following is completely wrong, has never happened, and if it did happen is fair, right and beneficial to society: At university of corporate headquarter state a start-up is formed by a professor, group of professors, and/or potentially a few students to commercialize their primarily NSF funded research. The university grants license to or transfers the IP registered and developed over the course of many years via their TLO in exchange for equity, trading talent and IP developed by that talent in exchange for a stake in the resulting company. A few years pass and an outside company, say a prominent tech company named after a fruit for example, buys out that company for relative peanuts(as demonstrated by say, a later lawsuit in which the NSF-funded IP alone was valued at billions) in order to acquire the talent and IP. They then use the acquired IP or license to sue the poo poo out of someone. For all of this the researchers get a relatively small payout, in the millions, the government gets nothing, the university gets a few million, the lawyers get many millions, the technology company named after a fruit gets awarded billions, and consumers of the resulting product ultimately foot the bill by inches for what basically comes down to research they had already payed for.
|
# ? Dec 10, 2013 22:30 |
|
Buffer posted:At university of corporate headquarter state a start-up is formed by a professor, group of professors, and/or potentially a few students to commercialize their primarily NSF funded research. The university grants license to or transfers the IP registered and developed over the course of many years via their TLO in exchange for equity, trading talent and IP developed by that talent in exchange for a stake in the resulting company. A few years pass and an outside company, say a prominent tech company named after a fruit for example, buys out that company for relative peanuts(as demonstrated by say, a later lawsuit in which the NSF-funded IP alone was valued at billions) in order to acquire the talent and IP. They then use the acquired IP or license to sue the poo poo out of someone. For all of this the researchers get a relatively small payout, in the millions, the government gets nothing, the university gets a few million, the lawyers get many millions, the technology company named after a fruit gets awarded billions, and consumers of the resulting product ultimately foot the bill by inches for what basically comes down to research they had already payed for. What's going on here is someone makes a bad business decision. Given the bad business decision hurt everyone involved, it clearly wasn't an attempt to screw the university or really had anything to do with the university.
|
# ? Dec 10, 2013 23:13 |
|
Alternately, the university (take your pick - well known ones include Harvard and UC - Berkeley) licenses their patent for assertion by another entity and pulls in a solid share of the profits from the license. The UC system pulled in around 125 million per year purely in licensing royalties over the past five years, not counting litigation revenue or an 87 million dollar prepayment of royalties for one patent from one licensee. The fact that some universities have chosen poorly in business doesn't mean the model is flawed as a whole.
|
# ? Dec 10, 2013 23:42 |
|
Those numbers sound like a lot, but the UC system is a $22 billion dollar operation, around $4 billion of which is government contracts and grants(source). Basically even when done well, the licensing revenue certainly doesn't hurt, but is pretty insignificant to the overall health of the institution. evilweasel posted:What's going on here is someone makes a bad business decision. Given the bad business decision hurt everyone involved, it clearly wasn't an attempt to screw the university or really had anything to do with the university. Nobody set out to screw anyone, malice isn't required for you to end up with a perverse outcome that hurt everyone you'd want the system to reward. And that's before you get into any consequences of requiring that largely non-profit research institutions, which are ostensibly doing the research it isn't business savvy for the private sector to do, also be business savvy.
|
# ? Dec 11, 2013 01:02 |
|
Buffer posted:Nobody set out to screw anyone, malice isn't required for you to end up with a perverse outcome that hurt everyone you'd want the system to reward. And that's before you get into any consequences of requiring that largely non-profit research institutions, which are ostensibly doing the research it isn't business savvy for the private sector to do, also be business savvy. Nothing you've described is the result of any IP laws. The bad outcome for the university is that someone sold an asset they had an interest in for pennies on the dollar. This is not a flaw of the IP system and nothing you've said remotely suggests it is. There's also absolutely nothing to suggest it was the university that was not business savvy: it is the people running the business who made the error. And they lost out quite a lot as well so like I said, nothing here is remotely the fault of IP or the University's actions.
|
# ? Dec 11, 2013 01:09 |
|
I think our views are not quite as far apart as I once did. Kalman posted:But that assumes the problem was already known and just hadn't been solved because no one had gotten around to it. Sometimes that's true, but your standard doesn't differentiate between obvious solutions to obvious problems and obvious solutions to non-obvious problems. If we assume that we also want to incentivize people whose invention consists of seeing the problem no one else had thought of before but that was easy to solve once thought of. A couple of examples from the PTO include the use of a sub coating to reduce degradation of omeprazole pills when no one realized that the uncoated version degraded, or reversing the sweep of a turbofan blade (easy for an ordinary turbofan designer to do) when there was no recognition that reversing sweep would improve a particular performance characteristic. The discovery that omenprazole pills degrade is (by which I really mean "should be") an unpatentable law of nature. Once that fact was discovered (which isn't patentable) the solution is obvious. I don't know the turbofan reference, but I see no immediate problem with that. It's the reversing the blade that's the patentable invention not the obvious means by which to do it, and it's useful because it improves performance. This situation is different than, for example, guessing that there will be driverless cars someday and so patenting driverless taxis: pre:1. A method and process of controlling a driverless taxi consisting of: a) a driverless car capable of navigating to a specified destination, b) a sequencing system capable of delivering specified destinations to the car 2. The method of claim one where the sequencing system a) first supplies a passenger's pickup location as a destination b) then supplies a passenger's destination location as a destination Kalman posted:There's also a huge issue in defining X for your standard. For example, if the "problem to be solved" is "create a public key crypto system" without the RSA patent, that seems hard. If the problem to be solved is "create a system where prime numbers define a key that can be provided to the public in order to encrypt something and define a second key that can be kept private for decryption", well, competent programmer might figure it out. Granted. I was trying to articulate what I believe the goal of the policy should be, not write the actual regulations. I could have been more clear on that point, and I don't know exactly how I would word the standard. I think your RSA comments support what I was arguing, though. The invention was that multiplying prime numbers a certain way will implement public key cryptography, not the specific code used. How to implement public key cryptography is a hard problem, how to do several things in a row is not. Kalman posted:Similarly, for the other patent, if the problem to be solved is "synchronize a group of different programs between two computers", you probably wind up with at least a couple of similar results to the patents management program which calls libraries corresponding to individual known programs to sync the data associated with those known programs (although some people will do it via offline file sync, some might do it by integrating libraries from the programs, some might do it by mirroring file writes, etc.). On the other hand, if the problem is "synchronize two computers" then you'll wind up with a bunch of different solutions, some of which might look like the patent, some of which won't. I think offline file sync and integrating libraries from the programs would be infringing. Mirroring file writes wouldn't work because we were talking about intermittent connections between the two computers.
|
# ? Dec 11, 2013 03:20 |
|
Kalman posted:But that assumes the problem was already known and just hadn't been solved because no one had gotten around to it. Sometimes that's true, but your standard doesn't differentiate between obvious solutions to obvious problems and obvious solutions to non-obvious problems. If we assume that we also want to incentivize people whose invention consists of seeing the problem no one else had thought of before but that was easy to solve once thought of. A couple of examples from the PTO include the use of a sub coating to reduce degradation of omeprazole pills when no one realized that the uncoated version degraded, or reversing the sweep of a turbofan blade (easy for an ordinary turbofan designer to do) when there was no recognition that reversing sweep would improve a particular performance characteristic. I'm not sure I believe in a non-obvious problem and I don't think either of those examples count as one. KernelSlanders mentioned the pill already, but I'd like to point out that, if the mechanical system isn't running at peak theoretical efficiency, that would automatically count as an obvious problem to be solved, so the possibly novel and non-obvious* approach that's patented in your example actually is attempting to solve the well-known problem that our machines aren't perfectly efficient yet. *I'm not qualified to even being to evaluate the obviousness or novelty of that approach quote:Similarly, for the other patent, if the problem to be solved is "synchronize a group of different programs between two computers", you probably wind up with at least a couple of similar results to the patents management program which calls libraries corresponding to individual known programs to sync the data associated with those known programs (although some people will do it via offline file sync, some might do it by integrating libraries from the programs, some might do it by mirroring file writes, etc.). On the other hand, if the problem is "synchronize two computers" then you'll wind up with a bunch of different solutions, some of which might look like the patent, some of which won't. It looks like they looked at how a sanely-designed database server synchronizes its databases and slapped 'handheld' all over it, which is a thing I've heard about happening in patents but with 'over a network' or 'on a mobile [computing] device' instead. "X thing that actually can be kind of hard (like synchronizing database records*) but more than once" - the 'do it again' hole - actually seems much worse to me because making the machine repeat a task is so basic to the field that it just feels so much more abusive. *I'd like to point out that even though I said that synchronizing databases can be kind of hard, I'm not sure it should be patentable but I'm personally of the opinion that software shouldn't be patentable, so take that as you will.
|
# ? Dec 12, 2013 00:20 |
|
From reading through this thread, I think the biggest disconnect for software patents is having an idea, or demonstrating it, that's kinda what I'm getting. So if you have 'novel idea' -> algorithm -> 'product' (figure out algorithm later) instead of 'this is exactly how you do this and it's unique and different'. I'll admit that I don't know much about the patent system even though I have one (US7539571 B2, my idea and implementation as an intern). It seems like the idea of patents has transitioned from products to ideas (in software). I think it should stay with products. If you want to patent X, make it first. Plinkey fucked around with this message at 09:27 on Dec 12, 2013 |
# ? Dec 12, 2013 09:23 |
|
Also patented: streaming videos that offer a view similar to a live classroom experience.quote:“...instructor at the head of the classroom with live-participants arranged between the instructor and the camera with a direct line of sight between the camera and the instructor allowing for the viewer participant to have unobstructed views while simultaneously allowing for the viewer participant to have live participants in the periphery, as if the viewer was attending a live class." quote:As you might remember from a couple months back, YogaGlo applied for two really similar patents. Initially, both were rejected by the U.S. Patent and Trademark Office (USPTO) based upon the existence of “prior art”—which is essentially evidence that the thing a patent is filed for already existed before said patent was filed. On October 7th, however, YogaGlo amended one of its previously rejected patent applications, adding the following caveat:That the camera used to record online yoga classes must "provide a participatory view [which means ‘a view observed by a participant in the rear of the class,’ according to YogaGlo] from a height of about three feet." Here's the claim in full: quote:
So it's pretty much just "making a yoga video in a studio, with a camera, with good production values, and then you record it or stream it onto the internet or something I guess, now give me money". This thing, this image right here? That's a novel innovation that's advancing the arts and sciences and is worthy of a monopoly on the concept for decades. There just seems to be this huge, gaping disconnect between the theoretical and the patents that actually get issued in real life. I don't have a problem with RSA being patented, I don't have a problem with LZW being patented, I accept most of the arguments in theory (except for design patents), but then in real life the patent office is rubberstamping patents for rounded corners and using a camera to film your yoga workout video. That seems like the real issue to me. You'd think for $1000 an hour (or whatever ridiculous number Kalman threw out there) the patent office could hire some people who have a reasonable understanding of practicing their art. But this one isn't even a technical thing where someone was trying to confuzzle the patent clerk. Have these clerks never seen a workout video before? Paul MaudDib fucked around with this message at 23:14 on Dec 12, 2013 |
# ? Dec 12, 2013 22:59 |
|
What? No, I get billed out at around 500 an hour, but I work for a law firm, not the PTO. Examiners get paid reasonably well but they're definitely not getting paid anywhere near as much as you seem to think. Also, fun fact: because they limited it to a three foot height by amendment, any height other than 3 feet is a-okay. Edit: also fun - that thing is difficult to infringe because of the shoddy draftsmanship. It claims the instructor, the position of the students, and the studio. Unless the studio is telling the students where to place themselves, the instructor where to place themself, and setting up the camera, there's a huge divided infringement issue (no single actor) and even the studio would only be liable for inducement liability, which requires strong intent to infringe, meaning that a good faith belief the patent is invalid is a defense. Kalman fucked around with this message at 23:17 on Dec 12, 2013 |
# ? Dec 12, 2013 23:13 |
Kalman posted:What? No, I get billed out at around 500 an hour, but I work for a law firm, not the PTO. Examiners get paid reasonably well but they're definitely not getting paid anywhere near as much as you seem to think.. A topped out GS14 examiner will make about $70 per hour.
|
|
# ? Dec 12, 2013 23:23 |
|
Shifty Pony posted:A topped out GS14 examiner will make about $70 per hour. I don't make 500 an hour (my firm does, but, well, capitalism). I make closer to 100 per hour billed. Then again, a topped out examiner will have been doing this a lot longer than I have (and yet still frequently be far worse at it than me...)
|
# ? Dec 12, 2013 23:27 |
|
Kalman posted:What? No, I get billed out at around 500 an hour, but I work for a law firm, not the PTO. Examiners get paid reasonably well but they're definitely not getting paid anywhere near as much as you seem to think. How is "about three feet" not vague? Is 3'6" infringing? What about 2'10" or 3' 0.03"?
|
# ? Dec 12, 2013 23:28 |
|
KernelSlanders posted:How is "about three feet" not vague? Is 3'6" infringing? What about 2'10" or 3' 0.03"? "In determining the range encompassed by the term about, one must consider the context of the term as it is used in the specification and claims of the application." If there's something in the spec that tells us what about three feet means, then we know what it means. If not, the claim is invalid as indefinite. (Amgen v Chugai, 927 f2d 1200.) E: based on a quick look at the spec, which only says "35 inches plus or minus enough not to change the perspective" I'd say there's a reasonable invalidity argument. Kalman fucked around with this message at 23:49 on Dec 12, 2013 |
# ? Dec 12, 2013 23:42 |
Kalman posted:I don't make 500 an hour (my firm does, but, well, capitalism). I make closer to 100 per hour billed. Then again, a topped out examiner will have been doing this a lot longer than I have (and yet still frequently be far worse at it than me...) Yeah, billed rates are strange, but then again so is the GS Pay system and quota system used by the PTO. Here's the Examiner pay scale: code:
The amount of work you have to do is governed by your "expectancy", which is set at the number of hours a GS-12 examiner is expected to take on all parts of an application, spread out between an initial search/action and a response to at least one amendment if that first action is a rejection. These expectancies vary anywhere from north of 30 hours to less than 10, supposedly in relation to how complex a technology is but they mainly haven't been tweaked since the 70s. The grades come with different adjustments: 0.5, 0.6. 0.75, 0.9, 1, 1.15 (1.25), 1.35. I could be off on the lower ones, I haven't had to worry about them for a while. At GS-13 the expectancy is split between those with partial signatory authority and those that still have to get all actions approved by a supervisor.
|
|
# ? Dec 13, 2013 00:00 |
|
I was surprised this thread wasn't about simply getting rid of patents (and potentially some additional IP laws to boot). In terms of best serving a social function, patents are terrible compared to direct funding. Ostensibly the advantage of patents is that private companies (and not the public) have to put up the (admittedly, not insignificant) initial capital outlay. Fair enough. But we do end up paying that cost one way or another as consumers - it has to be built into the price of something end users buy. The only "upside" of such a system compared to direct public funding is that only the people who make use of a particular patent end up paying for it. But most goods aren't luxury goods (it is a much smaller market compared to general consumer goods) so the vast majority of R&D funding that relies on patents is actually based on stuff everyone makes use of (or would like to). For the vast vast majority of people, that means that direct funding is always preferable to patents, and that's just from looking at the cost borne. There are many more drawbacks: Patent-funded R&D is always prioritized by profitability rather than social value - viagra will always get more funding then your mom's obscure cancer. Then there is also the profit itself - companies expect an actual return after all, not just breaking even (and that's after discounting all the R&D that ended up going nowhere). By far the worst drawback with patents though (and IMO it pretty much dwarfs all the other ones) is that patents just plain stifle innovation in a very direct way: you can't build on top of existing patents. Just think about that for a second, this should be a huge deal. All innovation happens on the shoulder of giants. Progress is best made iterating on existing solutions. As things stand if anyone wants to provide an incremental improvement over some drug, or gadget or technology, they either have to die of old age waiting for the patent to expire or pay the patent-holder's (rent-seeking) royalties. This is really obvious when it comes to software patents but it does apply to all patents - it's a massive barrier to innovation! IMO the same arguments can be made for copyright law as well. Is the social function to provide art? Entertainment? Distractions? Whatever it is, just pay for it directly. It is obvious to everyone what a bizarre and counterproductive effort DRM and copyright enforcement is. I mean here is something that benefits everyone that can be distributed to everyone free (or close enough to it), but no, instead of working on better ways of distributing such works we actually go out of our way to restrict it - it's ludicrous. "Oh but we have to support artists!", well yeah we sure do, so let's I dunno... do that. Again, it's not like we don't pay for them anyway, we just do it in the most roundabout way possible that also requires supporting a whole extra layer of distribution and rights-enforcers and returns for investors and so forth, oh and it also makes collaborative derivative work a non-starter. It's a terrible system. "Oh but this is just socialist day-dreaming, the established players will never in a million years let this happen!" I agree completely, but that goes for the whole thing. There is no point discussing what minor adjustment to the current fundamentally broken system might "work better" unless it "works better" for those existing established players, in which case, be sure that it's already being taken care of. Google, Apple, Microsoft, IBM etc know drat well software patents are bullshit, and they will continue their occasional skirmishes if they think they can come out ahead. But whenever they tire of this they sure as poo poo aren't going to start clamoring for the end of software patents, goodness no. They will just join up in patent federations that gives them mutual immunity while allowing them to come down hard on any uppity newcomer - you know, the way they already have been. The only rational improvement to the patent system is to get rid of it and replace it with direct funding. The only thing that will actually happen to the patent system is exactly nothing.
|
# ? Dec 13, 2013 00:06 |
|
Not that screeds in favor of central planning aren't interesting and all, but how do you plan to set the value of a given advance before, you know, you have the advance in hand? Particularly when it isn't necessarily going to be an advance that's predictable? Some things are amenable to a prize-funding approach, some are amenable to a direct funding for research approach, and some are actually best solved with a prospect (patent) approach. It depends on the characteristics of the problem. Known problems with relatively predictable values? Prize funding works well. Unknown problems with unknown values? Direct funding works as well as anything else, although prospect funding also works reasonably well. Known problems with unknown values? Prospect systems are ideal. (Unknown problems with known values don't exist for obvious reasons, so there is no fourth class to consider.) There are certainly flaws in the patent system and I'm all for increased public funding of basic research and prize funding of some things, but the idea that IP should all be directly funded assumes that a hard problem (valuation of IP) is easy.
|
# ? Dec 13, 2013 00:58 |
|
Even if it were a hard problem, what is the cost of getting it wrong compared to the public cost of patents as is? It would have to be hefty. I will grant that it is not a trivial, but I really don't know about hard. AFAIK most (i.e. the vast majority) of patents get filed by very large institutions, and they fund things pretty much the exact same way government research does: get some scientists, throw some money at them and hope for the best. Goals are set of course, but public organizations can do that just as well as private ones (and the goals will, of course, have public interests in mind rather then private ones). Some supervision and evaluation is required, but the major research schools already know how to do that anyway (it is in fact where private research draws all their staff from anyway, both actual researchers and their supervisors). Just give them money (as we already do with government research). Easy-peasy. Whatever patents get filed form tiny startups, well, for the most part that just doesn't happen because tiny startups don't have money for patent lawyers - best bet is they ask to be bought out by someone who does. They might ask for public research grants instead. Artists could be paid with traditional grants (as is already the case with stuff like public film boards and so forth) as well as some form of usage-based grants if you want to continue tying mass appeal to funding (there are pros and cons to this - I'd be happy to see both systems live along-side). We can estimate what people actually listen to / watch / play without DRM using polls, it need not be perfectly accurate. That's literally nothing more than what I pulled out of my rear end just now and I really think it would be just fine. A public policy economist could probably come up with something even better. I'm open to the idea that private involvement does do something that only really works in a competitive, decentralized environment, but I really think the onus should be on you to specify what that might be, just because in real life, in the world we actually live in, patent-funded R&D is not terribly competitive nor decentralized.
|
# ? Dec 13, 2013 01:20 |
|
Kalman posted:Not that screeds in favor of central planning aren't interesting and all, but how do you plan to set the value of a given advance before, you know, you have the advance in hand? Particularly when it isn't necessarily going to be an advance that's predictable? That's asking the wrong question. The patent system exists to promote the advance of technology (at least under American law which is what we're generally discussing here) rather than awarding the inventors the value of their invention through a quasi moral-rights theory. We do not need to worry about setting the value we award once the advance is made correctly, we need to worry about setting the incentives before the advance is made to make the advance correctly. Obviously, the post-discovery award is the incentive so they're somewhat linked but you've still got the wrong end.
|
# ? Dec 13, 2013 01:23 |
|
evilweasel posted:That's asking the wrong question. The patent system exists to promote the advance of technology (at least under American law which is what we're generally discussing here) rather than awarding the inventors the value of their invention through a quasi moral-rights theory. We do not need to worry about setting the value we award once the advance is made correctly, we need to worry about setting the incentives before the advance is made to make the advance correctly. Right. But if we're talking about a prize-type system, where we are offering some eventual prize for a given piece of technology, getting that number right is actually critical to the system functioning properly. If we get a pre-valuation number wrong in the patent system, it doesn't matter because that number has literally zero influence on whether someone pursues a technology or not (their own valuation estimate might, of course). If we get it wrong in a prize system, or in a direct funding system, it matters quite a bit because we will be underfunding or overfunding as a result. We don't have to engage in that in the patent system because we assume that the technology will self-value in the end and that that will promote the progress of technology in the aggregate by incentivizing inventors to disclose their discoveries in exchange for exclusivity for a limited time. Does it? Studies vary, and the results vary by field, but for the most part, it's not a drag on progress. Some fields respond better than others, of course. Pharma and semiconductors, as I said before, do very well in terms of responses to IP protection. High tech and software has a more mixed record. There are legitimate concerns in terms of patent quality, in medical access issues, in patenting of traditional knowledge derived inventions, etc., where a direct system does better, but then again, a direct funding system also has to be funded somehow, so the idea that "we pay more for our products because of patents so let's get rid of them in favor of direct funding" ignores the fact that the cost would then have to be borne in other ways. Kalman fucked around with this message at 01:33 on Dec 13, 2013 |
# ? Dec 13, 2013 01:25 |
|
Kalman posted:Actually, we file patents for tiny start ups all the time, and we are expensive. Smaller firms do patent filings for start ups far more frequently than we do. Almost all start ups file for protection in the start up phase because without it, they can't succeed - someone else will just copy their technology. That doesn't actually respond to "the vast majority" of patents are filed by large institutions. It's actually implicit in the statement that a large majority of patents are filed by large institutions (because that statement implies there's some filed by non-large institutions).
|
# ? Dec 13, 2013 01:26 |
|
Kalman posted:Right. But if we're talking about a prize-type system, where we are offering some eventual prize for a given piece of technology, getting that number right is actually critical to the system functioning properly. I'm sure some sort of X-Prize type contest may work well for certain problems, but it's not what I would base the whole thing on, no. quote:Does it? Studies vary, and the results vary by field, but for the most part, it's not a drag on progress. Some fields respond better than others, of course. Pharma and semiconductors, as I said before, do very well in terms of responses to IP protection. High tech and software has a more mixed record. quote:There are legitimate concerns in terms of patent quality, in medical access issues, in patenting of traditional knowledge derived inventions, etc., where a direct system does better, but then again, a direct funding system also has to be funded somehow, so the idea that "we pay more for our products because of patents so let's get rid of them in favor of direct funding" ignores the fact that the cost would then have to be borne in other ways.
|
# ? Dec 13, 2013 01:42 |
|
Mr. Wynand posted:Are we? I know I wasn't. I was talking about an upfront grant system, very similar to the existing government research grants around now (just larger). That's even worse. Not that there's no role for public grants (I rely on them for my salary), but abolishing all private sector R&D seems a bit extreme. A grant system requires a centralized institution (presumably some arm of the government) to administer it and pick which ideas should get funded. If my experience with the current grant system is any indication, moving all R&D to public funding is going to end in tears.
|
# ? Dec 13, 2013 03:24 |
|
KernelSlanders posted:That's even worse. Not that there's no role for public grants (I rely on them for my salary), but abolishing all private sector R&D seems a bit extreme. A grant system requires a centralized institution (presumably some arm of the government) to administer it and pick which ideas should get funded. If my experience with the current grant system is any indication, moving all R&D to public funding is going to end in tears. Why because public research has been such an unmitigated disaster? Private R&D is already highly centralized and receives no small amount of public funding in terms of special privileges and tax breaks to boot. Unless private R&D is somehow fundamentally different from the publically funded one, why wouldn't the existing and working public model work just as well? Seriously, do you have something against the NRC and related organizations? Point out something more specific then knee-jerk pro-market, anti-social-anything fear mongering.
|
# ? Dec 13, 2013 03:32 |
|
KernelSlanders posted:There doesn't have to be a One True Way, there just has to be an obvious way. Let's say you want to synchronize a bunch of programs simultaneously, how would you do that? Maybe you keep a list of all the programs you want synchronized and then synchronize each one in the list. The inventors on that patent didn't solve any hard problem, they just identified something people might one day want to do and patented the perfectly obvious way of doing it. I'm not anti-software patent per se. I have no problem with RSA. for example. That was a hard problem with a non-obvious solution. What I have a problem with is if they had claimed "an encrypted communication system that consists of a first terminal that receives a token from a second terminal which receives a token from the first terminal allowing it to decrypt the message." The standard should be, if you asked some random programmer how to solve problem X, with no knowledge of how the patent applicant solved problem X, would they come up with the same answer? If so, the answer is obvious. The thing is, if it's obvious, it's obvious, and we don't even get into patentable subject matter. To piggyback off another post: NJ Deac posted:Unfortunately we have a Supreme Court that is dead set on conflating obviousness into the subject matter eligibility analysis because there are some claims they just don't like but don't have the right art on record to reject them. This has resulted in a confusing, ambiguous, and unworkable standard that the Federal Circuit has been completely unable to apply consistently Bilski was terrible for patent law. It basically said "the claims are obvious, therefore it's not an invention" which would probably get me fired.
|
# ? Dec 13, 2013 03:51 |
|
WhiskeyJuvenile posted:The thing is, if it's obvious, it's obvious, and we don't even get into patentable subject matter. I can pretty much guarantee it wouldn't, at least if you dressed it up a little bit, based on the rejection I got yesterday. But yes, Bilski was a miserable decision that did nothing to help 101 be better understood. (In fairness, it's not really any worse than Federal Circuit case law, just bad in a different way.). CLS Bank will probably be more of the same. I'll have to find another word to add to the preamble of my Beauregard claims to make examiners happy and I might have to stop using one word or another in claims in favor of a new favorite word for defining software structures. Really, they just need to eliminate the notion that software is per se unpatentable, because some of it ought to be (the fact that claiming a machine that does something or a method of doing something is in theory patentable subject matter while software doing the same thing isn't is flat out stupid and an artifact of people taking the abstract idea doctrine out of the domain it belonged in, that of preemption of a field.). Software patents are generally bad patents because they engage in functional claiming and because software examiners are bad at obviousness research! not because software is somehow special and should be treated any differently.
|
# ? Dec 13, 2013 04:52 |
|
I used to give a poo poo about 101, but now it's just rote formula with respect to signals and software standing alone.
|
# ? Dec 13, 2013 15:00 |
I really wish the SC would do more to clarify the requirements of scope of enablement, claim interpretation, and 103 with respect to software in the manner of KSR instead of whatever the gently caress Bilski was.
|
|
# ? Dec 13, 2013 15:20 |
|
Kalman posted:Really, they just need to eliminate the notion that software is per se unpatentable, because some of it ought to be (the fact that claiming a machine that does something or a method of doing something is in theory patentable subject matter while software doing the same thing isn't is flat out stupid and an artifact of people taking the abstract idea doctrine out of the domain it belonged in, that of preemption of a field.). Software patents are generally bad patents because they engage in functional claiming and because software examiners are bad at obviousness research! not because software is somehow special and should be treated any differently. Would making software explicitly patentable eliminate the problem of patents being unreadable gibberish* or overly vague junk that's useless to actual software developers - the people other than the original inventor who are supposedly meant to benefit from their existence? Ideally, the patent would be (or read very much like) a published research paper that describes a novel algorithm. That's not as ideal as, say, a blog post with an implementation in your favorite language, but it's a hell of a lot better than the impenetrable crap I normally see in software patents and puts the papers in the public domain instead of behind paywalls. Software is special, by the way: it's applied math. If you're not comfortable with patents on mathematical concepts (like me!), then you might want to examine your comfort level with software patents. Could you lawyers see a clear way around a law that says something along the lines of "non-software patent claims against software systems are unenforceable"? Because I think I saw WhiskeyJuvenile claiming they could get around an explicit "software is not patentable" law, which I guess I can see happening given that the current system basically snuck the concept through a loophole, but if software were to become explicitly patentable, I'd still prefer all of the old patents to be disregarded because of the obviousness research problem. * The RSA patent actually is surprisingly readable, but I already know what RSA does and how it works, so that might be why. It's still making me doubt my position a little, though
|
# ? Dec 13, 2013 17:19 |
|
Kalman posted:If there's something in the spec that tells us what about three feet means, then we know what it means. If not, the claim is invalid as indefinite. (Amgen v Chugai, 927 f2d 1200.) EDIT: Or I guess some other yoga streaming companies could kickstart a reexam
|
# ? Dec 13, 2013 17:34 |
|
Everything in engineering is applied math/abstract concepts. That's kind of the point of engineering. Patents on rims for bicycle wheels are applications of physics. I am completely comfortable with patenting applied math so long as it's actually novel in the application and the application isn't an obvious outcome from the well understood math. You're basically engaging in the exact same problematic reasoning that Whiskey has attributed to Bilski - you have things you think are obvious and you are trying to use subject matter eligibility to get rid of them. I agree that "do it in software" shouldn't be patentable for a definition of "it" that was already known, but that's because it's obvious to implement a known process in software, not because software shouldn't be patentable. Software being made explicitly patentable would reduce some of the stupid boilerplate that gets put in, but wouldn't necessarily make claims more readable or more limiting. Software being rendered more unpatentable would force me to start writing claims directed to processors configured to do a process, or pure method claims, which would then be asserted against the software maker under inducement theories, so it wouldn't actually do much good at all.
|
# ? Dec 13, 2013 17:35 |
|
Kalman posted:Everything in engineering is applied math/abstract concepts. That's kind of the point of engineering. Patents on rims for bicycle wheels are applications of physics. I am completely comfortable with patenting applied math so long as it's actually novel in the application and the application isn't an obvious outcome from the well understood math. You're basically engaging in the exact same problematic reasoning that Whiskey has attributed to Bilski - you have things you think are obvious and you are trying to use subject matter eligibility to get rid of them. I agree that "do it in software" shouldn't be patentable for a definition of "it" that was already known, but that's because it's obvious to implement a known process in software, not because software shouldn't be patentable. Engineering actually makes physical objects that can be directly compared to each other, though. Software is just a set of data transformation instructions that could be represented in wildly different ways, even right on down to the machine-interpreted code. I'm think I understand what "functional claiming" means*, and if I'm right and it means something like making a claim on anything that takes as input X and produces Y as output: Y = X * 2, then I don't understand how you could talk about patenting software without doing that because all I have to do as a developer is Y = X << 1 to do the same thing in a different way. Yeah, that's a stupidly simple example, but the more complex and therefore less obvious a process is, the easier it'll be to write instructions that diverge from what's specified and get the same result, making it almost useless to pay someone thousands of dollars to patent that super specific invention. They already have an automatic copyright on their code, so what function does the patent serve if they aren't patenting "a function that multiplies a number by two"? If they can patent the function, I'm back to wanting to burn the whole thing down and if they can't I don't see why there should be a patent when there's already a copyright. Doubtless you'll say there's a middle ground, but apparently you could work around an outright ban on software patents, so how the hell can anyone trust the middle ground to hold when you've already said you can work around a solution that's the extreme opposite of what we have now instead of a watered down one? I should probably clear up that my ideal solution would be to make anything software causes a machine to do exempt from patent litigation. "Two machines with obvious structural differences that are caused by instructions stored in any medium that is mutable by design, to perform the same action or data transformation would not infringe" is what I'm coming up with for faux legalese - hopefully the intent is clear**. Reducing some of the boilerplate isn't encouraging. If the patent isn't usable by non-lawyers to re-create the invention, then the whole part of the argument for the patent system as a benefit to anyone but lawyers is just a bunch of hand-wavy bullshit and I go back to thinking it's a massive waste of effort that ought to just be abolished. *I read this: http://www.wired.com/opinion/2012/10/mark-lemley-functional-claiming/ which seems ** - everyone writing every law ever E: small edit to the pidgin legalese Munkeymon fucked around with this message at 22:55 on Dec 13, 2013 |
# ? Dec 13, 2013 21:12 |
|
Munkeymon posted:Engineering actually makes physical objects that can be directly compared to each other, though. Software is just a set of data transformation instructions that could be represented in wildly different ways, even right on down to the machine-interpreted code. A patent isn't on the specific physical object you invented. There's no real difference in patenting a chemical process or a software process, for example.
|
# ? Dec 13, 2013 21:49 |
|
Munkeymon posted:Engineering actually makes physical objects that can be directly compared to each other, though. Should we chuck EE patents as well? A lot of the stuff they can do is virtually indistinguishable from software as well.
|
# ? Dec 13, 2013 22:43 |
|
evilweasel posted:A patent isn't on the specific physical object you invented. There's no real difference in patenting a chemical process or a software process, for example. There is in reality, if not in law. According to this, you have to describe a novel apparatus to patent a new chemical process, but by the time they ruled on Diamond v. Diehr, general purpose CPUs had totally removed the need for a novel apparatus to run any given algorithm for decades. The ruling makes no sense what so ever in reality.
|
# ? Dec 13, 2013 22:43 |
|
Today I listened to a presentation at my work about how awesome it is that we have essentially "patent monopoly" on our industry, and that no new competitors can enter the market because of the patents that we control. To me, that seems hosed. I don't know enough about patent law to contribute much to this discussion, but there's a perfect example of patents stifling innovation and competition.
|
# ? Dec 13, 2013 22:47 |
|
|
# ? May 28, 2024 14:29 |
|
Toffile posted:Should we chuck EE patents as well? A lot of the stuff they can do is virtually indistinguishable from software as well. Can you cite an example because I can't think of anything that's hardware but acts like software.
|
# ? Dec 13, 2013 22:52 |