|
ASICs are the only thing that come to mind.
|
#
?
Dec 13, 2013 22:53
|
|
|
|
| # ? May 13, 2024 23:48 |
|
|
Munkeymon posted:Can you cite an example because I can't think of anything that's hardware but acts like software. Electronic engineering is basically just programming in a different medium at times.
|
|
#
?
Dec 13, 2013 23:03
|
|
|
evilweasel posted:Electronic engineering is basically just programming in a different medium at times. There's also a large crossover with advanced mathematics and systems engineering that doesn't translate well into hardware as well.
|
|
#
?
Dec 13, 2013 23:04
|
|
|
evilweasel posted:Electronic engineering is basically just programming in a different medium at times. But at the end of the day you're laying down a set physical representation of a circuit diagram that I can examine with a(n electron tunneling) microscope. That's at least a novel apparatus.
|
|
#
?
Dec 13, 2013 23:15
|
|
|
Munkeymon posted:But at the end of the day you're laying down a set physical representation of a circuit diagram that I can examine with a(n electron tunneling) microscope. That's at least a novel apparatus. How is different from your reasoning regarding software and general purpose CPUs? There are general purpose FPGAs or other PLDs.
|
|
#
?
Dec 13, 2013 23:22
|
|
|
Munkeymon posted:But at the end of the day you're laying down a set physical representation of a circuit diagram that I can examine with a(n electron tunneling) microscope. That's at least a novel apparatus. Patents are not limited to apparatus by statute. They can be directed to "process, machine, manufacture, or composition of matter." Your whole argument is based on an incorrect assumption. Again, what is the difference between a hardware 2 bit adder and software that implements the exact same functionality in code? (Of course, both are obvious at this point, but that's not the point of the statutory subject matter doctrine.) For that matter, if I write something in VHDL is it software or hardware? What if I simulate the VHDL on a computer, and then implement it into an ASIC? Why should those two things be treated any differently in practice?
|
|
#
?
Dec 13, 2013 23:25
|
|
|
Toffile posted:How is different from your reasoning regarding software and general purpose CPUs? There are general purpose FPGAs or other PLDs. Because you make a physical thing. I don't care what it does as long as you're not claiming to own the idea of performing the task it performs. I mentioned GPCPUs because the novel apparatus test apparently doesn't apply to a software process but does to a chemical one, which is strikes me as terribly inconsistent.
|
|
#
?
Dec 13, 2013 23:30
|
|
|
Munkeymon posted:Because you make a physical thing. I don't care what it does as long as you're not claiming to own the idea of performing the task it performs. I mentioned GPCPUs because the novel apparatus test apparently doesn't apply to a software process but does to a chemical one, which is strikes me as terribly inconsistent. It doesn't apply to either. Your understanding is wrong. Edit: to clarify, the novel part of the test doesn't matter to subject matter eligibility, and the apparatus part is a clue to eligibility but isn't a requirement. Just so we are all on the same page, to get a patent you need to pass a few different tests: 1). Are you trying to patent something that is of a type of thing that is patentable? (Subject matter eligibility, 101) 2). Is the thing you're trying to patent novel/not obvious? (Novelty/obviousness, 102 and 103.) 3). Did you provide adequate formal description and comply with the basic requirements for hoe you tell us what you know and what you claim to have invented? (Written description, enablement, definiteness, means plus function, and a few other minor doctrines like unclaimed essential matter. 112.) Just because something is valid subject matter doesn't mean it isn't obvious. Some of the problems in statutory subject matter case law come from courts trying to kill obvious patents without actually having the right art in front of them. The idea that software shouldn't be patentable is formally a 101 issue, but the arguments people make against software patents are often 102/103 arguments. Kalman fucked around with this message at 23:40 on Dec 13, 2013 |
|
#
?
Dec 13, 2013 23:34
|
|
|
Kalman posted:Patents are not limited to apparatus by statute. That's what I get for basing my legal understanding on reading wikipedia rather than going to law school, I guess  The difference is that the software is logically indistinguishable from a description of how to solve the problem of adding two binary numbers - it's merly a description of an algorithm in a specific language. Am I wrong that you can't patent algorithms? E: Does the court have a different understanding of "algorithm" than a dictionary? VHDL is equivalent to a circuit diagram in my mind, but I might be oversimplifying. Munkeymon fucked around with this message at 23:43 on Dec 13, 2013 |
|
#
?
Dec 13, 2013 23:40
|
|
|
Munkeymon posted:That's what I get for basing my legal understanding on reading wikipedia rather than going to law school, I guess It depends. Can you patent an algorithm? No. (gottschalk v Benson). Can you patent using an algorithm to solve a problem? Yes. (Diamond v Diehr). Defining when you're patenting the algorithm and when you're patenting solving a problem is the hard part. Which is why we've had Bilski, Prometheus, and soon CLS Bank.
|
|
#
?
Dec 13, 2013 23:53
|
|
|
Kalman posted:It depends. Can you patent an algorithm? No. (gottschalk v Benson). Can you patent using an algorithm to solve a problem? Yes. (Diamond v Diehr). Oh so that's what that means. I have a problem with that case, then, because using a rubber curing algorithm to cure rubber seems ridiculously loving obvious and I don't know a drat thing about curing rubber. The weird thing about this is that I feel like I might be coming around to being OK with short-ish patents on algorithms because a really novel, useful algorithm is something the inventor ought to get compensated for, but there's currently all this weird cruft around the concept and of course the total failure of obviousness research makes me wary of giving the patent system any more power than it has now.
|
|
#
?
Dec 14, 2013 00:06
|
|
|
Munkeymon posted:Because you make a physical thing. I don't care what it does as long as you're not claiming to own the idea of performing the task it performs. I mentioned GPCPUs because the novel apparatus test apparently doesn't apply to a software process but does to a chemical one, which is strikes me as terribly inconsistent. When I program an FPGA, I don't actually make a physical thing that is electrically different than something a GPCPU could produce. In the end, both a GPCPU and a FPGA are nothing more than a network of transistors that can be configured to form a physical circuit. At best, the major difference would be that as a more robust apparatus, a GPCPU is designed to handle faster switching between circuit designs. That doesn't seem like a terribly big difference. Munkeymon posted:Oh so that's what that means. I have a problem with that case, then, because using a rubber curing algorithm to cure rubber seems ridiculously loving obvious and I don't know a drat thing about curing rubber. Something is never novel when you strip out the thing that makes it novel. (Interfacing thermal sensors in order to compensate for a process design flaw)
|
|
#
?
Dec 14, 2013 00:23
|
|
|
Toffile posted:When I program an FPGA, I don't actually make a physical thing that is electrically different than something a GPCPU could produce. I know? I don't know what you think I'm arguing anymore, here so I'll try again: I think you should be able to patent novel, non-obvious configurations of logic gates wired together on a medium, but (probably) not the program/algorithm they run or any specific state of the gates.
|
|
#
?
Dec 14, 2013 00:59
|
|
|
Munkeymon posted:Oh so that's what that means. I have a problem with that case, then, because using a rubber curing algorithm to cure rubber seems ridiculously loving obvious and I don't know a drat thing about curing rubber. See? This thing you're doing here? This is a problem. We are talking about whether software should be patentable as a class, and you go "that software patent (that actually wasn't really a software patent) is obvious which is bad." That's what the Supreme Court does, so it's not like you're in bad company, but they are NOT related inquiries, and there's a reason for it. It's analytically sloppy to conflate the two. You can have something that's eligible, but obvious, and it isn't patentable. For example, aspirin - well known, but eligible. You can have something that's not eligible, but also isn't obvious. For example, a human being with the correct genetic sequences to walk on water. Not patentable (because of a specific exclusion on patents directed to human beings.) The point of subject matter eligibility isn't to eliminate obvious patents. It's to prevent patents that claim things we've decided, societally, simply should not have patents granted for policy reasons (NOT because a specific invention is obvious). Generally we bar things from patentability because they either sweep too broadly (the issue in Parker v Benson, where the patent would have covered any method of converting from binary to BCD) or because we have specific concerns (patenting people, or patenting methods that someone can carry out in their head or on scratch paper, although the later is sort of hard to define since in theory anything I can do on a computer I could do on scratch paper eventually). (Also, the Diehr patent is more than what you're making it out to be. For one thing, the Arrhenius Equation isn't a rubber curing algorithm; it's a general formula for the temperature dependence of reaction rates. For another, the patent claims not just the equation or method but a particular application of that method to automatically curing rubber in a press.)
|
|
#
?
Dec 14, 2013 01:09
|
|
|
Munkeymon posted:I know? I don't know what you think I'm arguing anymore, here so I'll try again: So you think you shouldn't be able to patent any processes at all, is what you're saying.
|
|
#
?
Dec 14, 2013 01:10
|
|
|
Kalman posted:See? This thing you're doing here? This is a problem. We are talking about whether software should be patentable as a class, and you go "that software patent (that actually wasn't really a software patent) is obvious which is bad." That's what the Supreme Court does, so it's not like you're in bad company, but they are NOT related inquiries, and there's a reason for it. It's analytically sloppy to conflate the two. That's why I said I had a problem with the case (as you described it). quote:(Also, the Diehr patent is more than what you're making it out to be. For one thing, the Arrhenius Equation isn't a rubber curing algorithm; it's a general formula for the temperature dependence of reaction rates. For another, the patent claims not just the equation or method but a particular application of that method to automatically curing rubber in a press.) Yeah, I read up and unsurprisingly it's more complicated than I thought. It looks like you couldn't put anything about using a computer system in your claims before that, right?
|
|
#
?
Dec 14, 2013 01:19
|
|
|
Kalman posted:So you think you shouldn't be able to patent any processes at all, is what you're saying. Well I'm a little less sure than I was a couple days ago, but a couple days ago I would have said "hell yes". I said earlier that a short patent lifetime on an algorithm doesn't sound too bad, etc, but I'm not qualified to fully evaluate the consequences of that kind of legal protection. I'm still confused as to how a 'process' is patentable but an 'algorithm' isn't even though the words effectively have the same meaning, but don't see why a chemical process and a data manipulation process are so hard to distinguish. How bad was my attempt at legalese back there? quote:"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" You can patent all the variations of 'but on a computer', 'but more than once' or 'but over a network' you want but it'd be useless to sue in most cases (except to be a dick).
|
|
#
?
Dec 14, 2013 01:40
|
|
|
... or we could just stop theorizing about the best way to cut off our own hands and replace patents with a direct funding model. 
|
|
#
?
Dec 14, 2013 02:11
|
|
|
Munkeymon posted:Well I'm a little less sure than I was a couple days ago, but a couple days ago I would have said "hell yes". I said earlier that a short patent lifetime on an algorithm doesn't sound too bad, etc, but I'm not qualified to fully evaluate the consequences of that kind of legal protection. I'm still confused as to how a 'process' is patentable but an 'algorithm' isn't even though the words effectively have the same meaning, but don't see why a chemical process and a data manipulation process are so hard to distinguish. Please articulate the reasons why you think the two should be distinguished from one another. I.e., why should a chemical process (add fluorine to water, apply mixture to steel) be more or less patentable than a data manipulation process (place data about fluorine into buffer named water, apply steel function to buffer)? In making this explanation, you are not allowed to refer to whether it would be obvious to do the thing or not. Additionally, as you explain this, consider whether your logic for why we should declare the data manipulation process not patentable applies to similar processes for hardware implementations that do the same thing. Also consider whether the argument applies if the patent is directed to a general purpose processor (a machine) running the data manipulation software, and to a general purpose processor controlling the chemical process. A chemical process and a data manipulation process aren't that hard to distinguish in general; there will, of course, be edge cases (e.g., the Diehr patent), but generally it would be possible to distinguish them if no one is trying to obfuscate. The question is why should we distinguish patentability based on this difference? It's not an easy argument to make, which is why courts screw it up all the time.
|
|
#
?
Dec 14, 2013 02:14
|
|
|
Mr. Wynand posted:... or we could just stop theorizing about the best way to cut off our own hands and replace patents with a direct funding model. Yes, that would be a very good way to cut off our own hands!
|
|
#
?
Dec 14, 2013 02:14
|
|
|
I think one of the key problems with software patents too is that they're frequently not necessary due to the obfuscating nature of software. The RAR format, for example, has made the decompression algorithm public as source code, with the caveat that the source code is not permitted to be incorporated into a compressor. Despite that head start and considerable popularity, to my knowledge, there still aren't any third-party compressors. There are plenty of proprietary products out there that do exactly the same thing, they simply refuse to disclose their secrets and the difficulty of reverse-engineering the software protects the commercial viability of them on their own, and they stimulate advancement via competition anyway.
|
|
#
?
Dec 14, 2013 04:18
|
|
|
Munkeymon posted:Yeah, I read up and unsurprisingly it's more complicated than I thought. It looks like you couldn't put anything about using a computer system in your claims before that, right? You could, it just was a grey area until the supreme court decision. I seem to recall at least one of the Justices musing in the Bilski arguments that a newly programmed computer wasn't any different from the old machine, but that it was just a process disguised as an apparatus claim. Ah, found it, with the memorable end quip... quote:JUSTICE STEVENS: I don't understand why that isn't just the application of a process, which -- which is not itself patentable subject matter, to a particular machine that can use the process -- Mr. Stewart was representing the US Patent Office/Government in the case. Whether or not a general purpose computer programmed with a program becomes a particular machine is something we really need a ruling on, but goodness will it be a clusterfuck if the lines between 101 and 102/103 isn't clear. I know that reading anything into Justice questions in oral arguments is very unreliable but the reasoning in those kind of questions make me worry that we'll get another Bilski.
|
|
#
?
Dec 14, 2013 04:21
|
|
|
Kalman posted:Yes, that would be a very good way to cut off our own hands! You know, considering you've not actually bothered to respond to me, I have to admit, this sort of glib reply kind of irks me. OTOH, I don't know if it's even fair to expect a patent lawyer to admit patent law is useless - I mean I don't actually think I'm a big enough man that I could admit my day job is ultimately useless. It may well be, but I don't think I could actually remove my personal bias from the argument even if I tried. Still, I was expecting something a little more substantial than a strawman (even if unintentional) and the usual appeals to The Free Market Is The Only Way To Solve Basic Allocation Problems.
|
|
#
?
Dec 14, 2013 04:46
|
|
|
Given that you haven't bothered to respond to basic criticisms of the idea that direct funding is somehow flawless, a little bit of poking fun is deserved, I think. That said: do you not see the basic problem in centralizing research funding in a political structure? Particularly when you're also proposing abolishing any reward structure for people who aren't granted research funding? I mean, as an example I'm quite familiar with: it took Harold Black, the guy who came up with the concept of linearization of amplifiers via negative feedback (an incredible insight which is fundamental to modern electronics) almost a decade to convince the Patent Office that his invention was worthwhile. That's just one example of how centralized assessment of the value of research is problematic even when there's good intentions on everyone's part. Of course, people are people. That's why, even though RCA engineers had a functional LCD in the 60s, that project was killed off because of personal dynamics on the project teams. So RCA lost out on a huge product line to companies like TI. Decisions get made not just on technology but on personality. Good thing some of those researchers went to other startup that did believe in LCDs. Now add in the current political dynamic. Think about how hard it is to research stem cell therapies, birth control, and similar controversial things right now due to politics. Imagine that the current American political system controls all research funding (in practice, as private funding is going to be significantly reduced if everything can be legally copied and if they can just rely on public results to base products on.). Are you still comfortable with a direct grant system? There's a place for direct funding of research, but that place isn't as the only type of research structure. Kalman fucked around with this message at 05:20 on Dec 14, 2013 |
|
#
?
Dec 14, 2013 05:12
|
|
|
Kalman posted:Given that you haven't bothered to respond to basic criticisms of the idea that direct funding is somehow flawless, a little bit of poking fun is deserved, I think. quote:That said: do you not see the basic problem in centralizing research funding in a political structure? Particularly when you're also proposing abolishing any reward structure for people who aren't granted research funding? quote:Now add in the current political dynamic. Think about how hard it is to research stem cell therapies, birth control, and similar controversial things right now due to politics. Well, technically, the American political system can control just about anything if it wants to. Bush merely banned public funding for stem cells but he could just as easily have outright banned it if he had the mandate/support for it. Ok so, granted, the fact that public funding for research is limited in scope gave him a handy avenue for a compromise, but it's not like he couldn't have found a different one if he only had an all-encompassing system to work with. E.g. if he had cut all funding for stem-cell research it would be the same as if he had banned it, outcry and all. Having said all this, look, I do admit there are problems with centralizing R&D this way - but I think they are actually quite subtle. The problems with patents otoh are, relatively speaking, quite obvious, and more importantly, they are immediate and pressing. Patents are seriously hampering development in medicine and software and that's just the stuff that's obvious to everyone. To argue against myself some more: this is a big change, and big changes are always dangerous because they can have such unpredictable effects. So I am definitely open to ideas about how this can all go terribly, terribly wrong. So far I'm only getting weaksauce though.
|
|
#
?
Dec 14, 2013 05:57
|
|
|
Problems in the current system are always more pressing than problems in the system that isn't implemented yet. If you want to make massive changes in the way essentially all research in the US functions, I think you need to come up with more than "we'll our current systems problems are here right now." Not to mention, your arguments for why it would be an improvement don't match up with reality. For example: patents aren't generally hindering developments in medicine. Pharma is one of the areas where the economic theory aligns very well with the practice such that patents do drive pharmaceutical innovation. Even entities which are generally critical of patent incentives tend to hold up pharma as an example of an industry where patents work well (e.g., the NAS report from a couple years ago.) Patents retard access to those developments for a period of time, but that's a very different issue. Best evidence is that without a patent system, we'd have fewer and worse drugs, since the innovation cost is very high and the copying cost is very low. You can see examples of how the law has worked around this in medicines in the ANDA system, where once a drug is eligible for genericization (which can piggyback on the expensive human safety studies done by the brand name drug) we still provide a more limited for of exclusivity to the first generic company to challenge the brand name patent. Without the exclusivity period, you'd actually wind up with no one challenging the brand name patent since they can benefit more by free riding on the first challengers work. It's a pretty basic free rider problem. Similarly, in a hypothetical direct-only system, you have public funding of research into new drugs and into their human safety (which will be funded by increased taxes, so it wont actually make pharma research cheaper.). At the end of the day, you're going to have the perverse incentive that no one wants to be the first person to figure out how to commercialize the drug because they won't be able to protect the manufacturing processes they come up with from copying by competitors. If you publicly fund that research, then sure, anyone can use the manufacturing process, but you're going to wind up with a race to the bottom in terms of cutting manufacturing costs, since thats the only avenue for competitive differentiation left (well, that and advertising). So we have a race to cut wages and try to advertise your product over everyone else's, or else we have public manufacture to avoid the arms race. At that point you are essentially a centrally planned economy with all of the flaws that that entails. I'm curious what you think of http://scopeblog.stanford.edu/2012/12/05/nih-funding-mechanism-totally-broken-says-stanford-researcher/ which pretty much says the public funding model for medical research is broken.
|
|
#
?
Dec 14, 2013 06:29
|
|
|
Kalman posted:Problems in the current system are always more pressing than problems in the system that isn't implemented yet. If you want to make massive changes in the way essentially all research in the US functions, I think you need to come up with more than "we'll our current systems problems are here right now." quote:Not to mention, your arguments for why it would be an improvement don't match up with reality. Man, you know the rest of the world is absolutely astounded at how much Americans pay for their drugs right? Of course the costs are huge, but again, what makes you think you couldn't get the same results directly funding the same amount of R&D work? Again, the only difference is, do you pay it evenly based on taxes, or by playing insurance-company bingo? And even in the best case scenario where nobody is being dicked around for coverage for some cutting-edge drug they need, you still have no choice but to also pay for marketing, the drug company's profits, the insurance company's profits, the patent filings etc etc. And if that drug could actually be made available for peanuts? Too bad, you're paying full rent on that until the patent expires or generics kick in. quote:Similarly, in a hypothetical direct-only system, you have public funding of research into new drugs and into their human safety (which will be funded by increased taxes, so it wont actually make pharma research cheaper.). At the end of the day, you're going to have the perverse incentive that no one wants to be the first person to figure out how to commercialize the drug because they won't be able to protect the manufacturing processes they come up with from copying by competitors If you publicly fund that research, then sure, anyone can use the manufacturing process, but you're going to wind up with a race to the bottom in terms of cutting manufacturing costs, since thats the only avenue for competitive differentiation left quote:So we have a race to cut wages and try to advertise your product over everyone else's, or else we have public manufacture to avoid the arms race. quote:At that point you are essentially a centrally planned economy with all of the flaws that that entails. quote:I'm curious what you think of http://scopeblog.stanford.edu/2012/12/05/nih-funding-mechanism-totally-broken-says-stanford-researcher/ which pretty much says the public funding model for medical research is broken. Now this is the sort of thing I was looking for. Definitely a valid concern, as is cronyism in general. But is it intractable, or inherent to the system? I don't see why. Your own article suggests a solution: base some of the funding on objectively measurable citation impact. We can also go out of our way to "decentralize" the decision making bodies. Allow several of them to operate independently of each-other. This would actually come quite natural scaling something like the NIH to a nation the size of the US. This also isn't exactly a non-issue to existing for-profit research companies. We can look at whatever problems they ran into and their solutions, as well as of course other nations' public research institutions. This isn't something that has to be done going in blind.
|
|
#
?
Dec 14, 2013 07:48
|
|
|
Ugh, long post written. It was too long to be readable. The short version: Many of the problems you identify (e.g., insurer profit margins, drug advertising) are not actually issues with the patent system and would not be fixed by public funding of research. Drug advertising (which I would love to see banned) would probably be made worse, as it would be one of the few ways that companies could differentiate themselves (since they couldn't differentiate on technology, since it would be perfectly legal to copy your competitor's advances in manufacturing.) Patent costs are a miniscule portion of drug costs. Patent filing might be 100k, a full litigation (not settled) is 5-20m. Clinical trials alone are hundreds of millions of dollars. So eliminating the patent system reduces drug costs by not very much, outside of the removal of the monopoly rent that patents currently convey. Drug company profits are also not entirely patent-driven. Large brand name pharma companies run at around 15-25% profit margins. Large generic companies run at 10-15% profit margins. From a quick look, generics spend roughly as much (percentage of revenue-wise) on non-R&D costs such as marketing and administration (25%) and cost of manufacture (40-50%) as brand name makers. They spend roughly half as much on R&D (~7% vs ~15%). Generics actually do a decent amount of manufacturing/processing R&D, and do still have approval costs to deal with, although those are much lower, so its not surprising to see a decent chunk of R&D spending by them. Since the major difference between a generic company and a branded company is that one does patented development and the other does follow-on work, we can attribute most of this difference in profit margin to drug patents. So of the 5-15% increased profit margin for branded over generic, ~7-8% of it is recovery of the increased R&D cost. We're looking at a patent rent of approximately between -2% and +7% (back of envelope estimate) on drugs. In other words: drug patents don't seem to be the cause of the cost problems in healthcare. I remain unconvinced that the patent system is a drag on either pharma innovation or on treatment outcomes. In high tech, it's a bit different, but high tech is generally even less suitable to public funding, since the patent landscape in high tech tends to be much broader and valuation is almost purely driven by end user preference. As hard as valuing a drug before testing is, it's even harder to figure out how much a given new electronics advance will be worth in the end. (See, e.g., the Internet.) So, in exchange for large high tech companies not getting stuck with patent holdup issues and startups being allowed to create technology unfettered (which large industries will promptly take without any compensation, since now they're allowed to), you want to effectively convert every industry into a commoditized industry except for our publicly funded research sector. Commoditized industries have worked out incredibly well for employees, after all - the American television, cellphone, and semiconductor manufacturing industries are just going gangbusters. Commoditization works out well for consumers, but it's hard to be a consumer when you're working minimum wage jobs. I'm not sold.
|
|
#
?
Dec 14, 2013 10:33
|
|
|
Kalman posted:Imagine that the current American political system controls all research funding (in practice, as private funding is going to be significantly reduced if everything can be legally copied and if they can just rely on public results to base products on.). Are you still comfortable with a direct grant system? I have no idea what you're talking about, the NSF grant system is generally viewed to be working excellently, with the largest problem being not enough money to give out.
|
|
#
?
Dec 14, 2013 17:18
|
|
|
Munkeymon posted:Because you make a physical thing. I don't care what it does as long as you're not claiming to own the idea of performing the task it performs. I mentioned GPCPUs because the novel apparatus test apparently doesn't apply to a software process but does to a chemical one, which is strikes me as terribly inconsistent. Munkeymon posted:Well I'm a little less sure than I was a couple days ago, but a couple days ago I would have said "hell yes". I said earlier that a short patent lifetime on an algorithm doesn't sound too bad, etc, but I'm not qualified to fully evaluate the consequences of that kind of legal protection. I'm still confused as to how a 'process' is patentable but an 'algorithm' isn't even though the words effectively have the same meaning, but don't see why a chemical process and a data manipulation process are so hard to distinguish. To simplify things a bit, a process done on physical things should be patentable for the same reason that physical things are themselves patentable. The rub is that an algorithm is a process/recipe carried out on numerical objects. You can't get a patent on an abstract object, so it's inconsistent to have them for recipes carried out on abstract objects. I'll agree with the first thing you said: the idea of performing a task should probably not be patentable. In other words, a process which results in cured rubber should be patentable. "A Process which Results in Cured Rubber." [Period] should not be. Some algorithms may be owned to an extent by copyright. If someone built a "D" compiler which swapped around a bunch of keywords in the C language, then transcribed C programs written for the PC and compiled them on a MAC, then that would surely be a copyright violation, even though neither the source code nor compiled version would be the same. So using the same algorithm would veer close to copyright infringement, depending on how flexible you are with sameness. Phyzzle fucked around with this message at 23:12 on Dec 14, 2013 |
|
#
?
Dec 14, 2013 19:50
|
|
|
Kalman posted:Ugh, long post written. It was too long to be readable. Ok but 20m out of 100m in trials coast is still, you know 20%. Let's say 2-5% for that. 10% for the additional profit margin for non-generics. 2-7% the patent rent. So we still end up with 14-22% savings which isn't world changing, but nothing to sneeze at. I'm not sure how this jives with real world generics vs brand name costs though - like, the price difference I actually see going to the pharmacy is on the order of 1000% (literally, 10x more expensive, sometimes more). What am I missing here? Also please keep in mind that I believe having immediate access to new inventions and the ability to perform incremental development is the most valuable part here, rather than just straight up cost. quote:In high tech, it's a bit different, but high tech is generally even less suitable to public funding, since the patent landscape in high tech tends to be much broader and valuation is almost purely driven by end user preference. As hard as valuing a drug before testing is, it's even harder to figure out how much a given new electronics advance will be worth in the end. (See, e.g., the Internet.) quote:So, in exchange for large high tech companies not getting stuck with patent holdup issues and startups being allowed to create technology unfettered (which large industries will promptly take without any compensation, since now they're allowed to), you want to effectively convert every industry into a commoditized industry except for our publicly funded research sector. Commoditized industries have worked out incredibly well for employees, after all - the American television, cellphone, and semiconductor manufacturing industries are just going gangbusters. Commoditization works out well for consumers, but it's hard to be a consumer when you're working minimum wage jobs. You've not really explained why the gently caress wage compensation has dick-all to do the profitability in that sector (hint: it doesn't). Video games are a cool high-tech industry posting record profits. They treat their employs only slightly better than disposable garbage. I wonder if it actually has nothing to do with margins and everything to do with good ol' supply & demand (too many kids want to grow up to "make video games", and are thus plentiful and infinitely replaceable). Compare to: any other kind of programming, which pays very well, because it ain't nearly as sexy and demand outstrips supply. American labor is without a doubt suffering, and I will be the first in line to agree with that, but it has absolutely nothing to do with commoditizing and everything to do with the collapse of unions.
|
|
#
?
Dec 14, 2013 22:30
|
|
|
Mr. Wynand posted:Ok but 20m out of 100m in trials coast is still, you know 20%. Let's say 2-5% for that. 10% for the additional profit margin for non-generics. 2-7% the patent rent. So we still end up with 14-22% savings which isn't world changing, but nothing to sneeze at. I'm not sure how this jives with real world generics vs brand name costs though - like, the price difference I actually see going to the pharmacy is on the order of 1000% (literally, 10x more expensive, sometimes more). What am I missing here? Generics are priced on the marginal cost (ish), non generics are still recovering research costs. Because a branded drug is only really a branded drug for 5-10 years most of the time, the recovery occurs over a short period of time, while generics sell for a longer period of time. Remember also that my estimate was for portion of drug company profits due to patent, not for cost of drugs. Cost of drugs is a different analysis - if you assume free r&d then yes, drugs will be cheaper, but as we already discussed, that's just because you're paying for the development in taxes, so the cost is simply shifting and isn't actually due to patents. Trial costs aren't the majority of drug development costs. Estimates vary, but simply taking number of approved new chemical entities and dividing that by total r&d (which has flaws as a methodology) gives a current NCE cost (with the NCE entirely subsidizing improved entities and similar follow on research) of around 4 billion dollars per drug. So the cost of the patent system in terms of system costs is around a half percent, not the 2-5% you pulled out of your rear end. quote:Also please keep in mind that I believe having immediate access to new inventions and the ability to perform incremental development is the most valuable part here, rather than just straight up cost. You have immediate access to new inventions and the ability to perform incremental development in the patent system. You just have access controlled by the inventor and can't sell your incremental development unless they let you (of course, they can't sell your incremental development either.). This actually has positive effects as well as some negative ones, including on how incremental development is performed - there's an article I link at the end of this post that outlines some of these. quote:I feel like you keep talking past me on this: I really don't think you would do this via after-the-fact prizes, you just pay the research up front, same as the NRC and same as private companies. Remember, the government isn't paying people to invent iPhones, merely the patented bits of the iPhone: lithium-ion batteries, gorilla glass, advanced power management operating systems, application security jails (i'm assuming we can all agree bullshit iPhone patents aren't worth considering here - e.g. rounded corners or whatever the latest most ridiculous filing-du-jour is). The major tech companies do *some* of this but it is a very minor part of their existence. With some famously notable exceptions (Xerox PARC), most major theoretical advances are made at research universities by PhDs. Implementation is of course still a tremendously difficult job, and that is what established tech companies as well as plucky garage startups will continue to do - and certainly for the latter group, patents are nothing but huge source of business risk they can ill afford. I wasn't talking about prizes, I was talking about your model. You "pay the research up front". Assume scarcity (safe assumption). We now need some way to evaluate what things deserve what funding. Hence my statement that you do need to have some way to evaluate things. Your proposal now seems to be that, what, public funding will do basic research and companies will do implementation research? Why would they, when they can take the public basic research and wait to copy the next guys implementation research? Why would a startup bother starting up if IBM can take their idea and commercialize it (something that actually is a huge problem, and the reason a lot of startups seek patent protection in the first place/the reason VCs won't invest in them without it.). And if you propose some reason they can't copy it, how is it any different from the current system? I'm still waiting for a response to any of these criticisms. I feel like you don't understand the basic arguments FOR patents and are listening exclusively to the horror stories in the press, so I am going to link you to one of the better basic articles on how the patent system functions. http://bayhdolecentral.com/bremmerPDF/The_Nature_and_Function_of_the_Patent_System_by_Edmund_W._Kitch.pdf -- there is a ton of back and forth on Kitch's article, but it will at least give you the classical perspective on patent justifications beyond what the popular conception is. Kalman fucked around with this message at 22:52 on Dec 14, 2013 |
|
#
?
Dec 14, 2013 22:50
|
|
|
Kalman posted:Generics are priced on the marginal cost (ish), non generics are still recovering research costs. Because a branded drug is only really a branded drug for 5-10 years most of the time, the recovery occurs over a short period of time, while generics sell for a longer period of time. Remember also that my estimate was for portion of drug company profits due to patent, not for cost of drugs. Cost of drugs is a different analysis - if you assume free r&d then yes, drugs will be cheaper, but as we already discussed, that's just because you're paying for the development in taxes, so the cost is simply shifting and isn't actually due to patents. Oh ok - I think I see your point. Development costs (private or public) would have to equal the current revenue (even at brand-name, exclusive-patent prices), minus profit margin, marketing expenses (though this can/should be dealt with separately regardless of what we do with patents), minus the difference in manufacturing costs between generics and brand manufacturers (assuming *all* drug manufacturers under the proposed system happens the way generics do now... which seems reasonable). Brand-name drugs are orders of magnitude more expensive up front because they have to collect *all* their R&D costs back before their exclusive patent expires. An all-public, all-generic system would collect the exact same amount (except the aforementioned differences in profit margin and manufacturing cost), just spread out differently. Makes sense. So, aforementioned difference would be closer to, say, 15% or so by your reckoning? Well, I could live with that.. I mean I never expected it to be larger than maybe 40% even in the worst-case scenario. I do agree that such paltry savings alone would not be worth turning the R&D world upside down for. quote:You have immediate access to new inventions and the ability to perform incremental development in the patent system. You just have access controlled by the inventor and can't sell your incremental development unless they let you (of course, they can't sell your incremental development either.). This actually has positive effects as well as some negative ones, including on how incremental development is performed - there's an article I link at the end of this post that outlines some of these. Well these are the more negatives, as I've said from the beginning. Thing is those license fees are always going to be either rent seeking (which is pretty stifling) or outright not available, because more of then than not, no amount of royalties are worth having a huge competitive advantage in your market. quote:I wasn't talking about prizes, I was talking about your model. You "pay the research up front". quote:Your proposal now seems to be that, what, public funding will do basic research and companies will do implementation research? As things stand, if you have a business plan and that business plan relies on being able to press a patent claim against much large companies, most VCs won't touch that poo poo with a ten foot pole. VCs definitely like patents because they make you a much tastier buy-out target (and everyone likes being bought out) and serve as at least partial deterrent when the patent trolls come calling, but they definitely don't want you going out picking fights over patents as a negative-cashflow startup. In general, if your business plan is such that it can simply be copied by a larger company, that's what's referred to as "picking up pennies in front of the steamroller". E.g. you can make some tremendously useful plugin for facebook or office or something, but if it really is that useful, one day Facebook or Microsoft are just about guaranteed to simply implement it in their base product, and hopefully you made your investment back by then or they were nice enough to buy you out to save themselves the implementation costs, but that's about it in terms of long term strategies, because "fighting it with patents" is not seen as a realistic option. Tl;dr: You already have to work around this, nobody counts on patents helping them out. quote:I feel like you don't understand the basic arguments FOR patents and are listening exclusively to the horror stories in the press, so I am going to link you to one of the better basic articles on how the patent system functions. http://bayhdolecentral.com/bremmerPDF/The_Nature_and_Function_of_the_Patent_System_by_Edmund_W._Kitch.pdf -- there is a ton of back and forth on Kitch's article, but it will at least give you the classical perspective on patent justifications beyond what the popular conception is. I don't think I'm that unfamiliar with how the patent system is supposed to work, or why free-markets are supposed to be, generally (in very specific, not actually that common circumstances), efficient. I should maybe mention that I (along with some remarkably large swaths of D&D) spent a fairly long time in the libertarian camp before "crossing the isle" so to speak, so I am at the least somewhat familiar with efficient-market-hypothesis rhetoric and theory. Looks like a nice enough read though - I've added it to my queue.
|
|
#
?
Dec 14, 2013 23:31
|
|
|
Kalman posted:(which large industries will promptly take without any compensation, since now they're allowed to)
|
|
#
?
Dec 15, 2013 01:28
|
|
|
OneEightHundred posted:Like I said, this isn't even true because of the enormous reverse-engineering effort required to "take" an algorithm from software that's been compiled into machine code or exists as a black box at the other end of a network connection. Small companies license out proprietary, unpatented software to larger players all the time. That is just because the development costs are higher than the license costs. The small company in your example, profits by having more than one licensee. I don't think anyone's arguing that eliminating software patents would kill the software industry, but rather that there are some innovations deserving of patents that are easily copied. Moreover, I haven't seen any compelling argument for domain exclusion of software generally, just a lot of terrible software patents. Reform should be addressed at the core problem: how do we stop the really bad patents (which aren't limited to software by any means) while permitting the good ones? Do we even agree on which are which?
|
|
#
?
Dec 16, 2013 16:25
|
|
|
Kalman posted:Please articulate the reasons why you think the two should be distinguished from one another. I.e., why should a chemical process (add fluorine to water, apply mixture to steel) be more or less patentable than a data manipulation process (place data about fluorine into buffer named water, apply steel function to buffer)? In making this explanation, you are not allowed to refer to whether it would be obvious to do the thing or not. Additionally, as you explain this, consider whether your logic for why we should declare the data manipulation process not patentable applies to similar processes for hardware implementations that do the same thing. Also consider whether the argument applies if the patent is directed to a general purpose processor (a machine) running the data manipulation software, and to a general purpose processor controlling the chemical process. Phyzzle articulated it better than I would have - you can't patent the abstract and, until a hardware implementation is involved, every algorithm is abstract. I'm also concerned about the societal cost of patentable algorithms. Compared to spinning up a new chemical production line or building a water treatment plant, deploying software to a bunch of computers is essentially free and instantaneous, so you can very easily end up with a situation where millions of people are infringing on thousands of patents simultaneously. I think we'd have to convert most of the matter in the solar system to lawyers to work that legal mess out - or we could agree that it's not worth making the mess to begin with.
|
|
#
?
Dec 16, 2013 17:51
|
|
|
KernelSlanders posted:That is just because the development costs are higher than the license costs. The small company in your example, profits by having more than one licensee. I don't think anyone's arguing that eliminating software patents would kill the software industry, but rather that there are some innovations deserving of patents that are easily copied. Moreover, I haven't seen any compelling argument for domain exclusion of software generally, just a lot of terrible software patents. Reform should be addressed at the core problem: how do we stop the really bad patents (which aren't limited to software by any means) while permitting the good ones? Do we even agree on which are which? Anything that's easier/cheaper to copy than license is going to be too obvious or trivial to (deserve a) patent, at least for the foreseeable future. Maybe this only becomes clear when you've been developing software for a few years, but it's the state of the industry and until it changes, copyright is sufficient protection. E: maybe the answer is to require patent examiners to have at least five years of actual software development experience 
Munkeymon fucked around with this message at 20:01 on Dec 16, 2013 |
|
#
?
Dec 16, 2013 18:00
|
|
|
Munkeymon posted:Anything that's easier/cheaper to copy than license is going to be too obvious or trivial to (deserve a) patent, at least for the foreseeable future. Maybe this only becomes clear when you've been developing software for a few years, but it's the state of the industry and until it changes, copyright is sufficient protection. Right, the patent isn't there to protect you from your customers; it's there to protect you from your competition. If you have a truly innovative piece of software the patent prevents some other software company from piggybacking off your development costs and licensing its version of your product to your customers. If that were permitted, it would be nearly impossible to recoup development costs -- particularly if the hard part isn't writing the code, it's figuring out what application to write.
|
|
#
?
Dec 16, 2013 21:05
|
|
|
KernelSlanders posted:Right, the patent isn't there to protect you from your customers; it's there to protect you from your competition. If you have a truly innovative piece of software the patent prevents some other software company from piggybacking off your development costs and licensing its version of your product to your customers. If that were permitted, it would be nearly impossible to recoup development costs -- particularly if the hard part isn't writing the code, it's figuring out what application to write. Except that your competitors can only maybe piggyback off of your market research* and are going to be years behind you in development time and while they're catching up, your hot new product becomes the brand name most associated with the product you've invented. Now you have a stable business or yet another stable product that, sure, other people are going to try to compete with, but that's what we as a society want to happen. We don't want anyone to have a twenty year monopoly on, say, 'electronic delivery of textual and/or graphical mail' or 'a game where you shoot mans with a gun from a first-person perspective'. Think of any web site that hasn't been updated in 10 years and imagine the whole internet was like that because  we don't need to update it our customers seem ok with it (because they have no alternative to compare it to of course) Altavista/Yahoo's hosted Babelfish service looked and worked substantially the same for a long drat time: http://web.archive.org/web/20110408105943/http://babelfish.yahoo.com/ Then Google came along and made a better one: http://web.archive.org/web/20090514144436/http://translate.google.com/translate_t Now Babelfish is doing this: http://www.babelfish.com/ and Microsoft has theirs to compete with Google. Three products that do the same extremely complicated thing, probably very differently, forcing each other to get better. They are, as far as I can find, all independently built. *this is extremely common in every industry! Ever wonder why coffee shops cluster? One chain does the market research to find a good location and the others all follow them into that market. Works the same with kids toys, clothes and Hollywood movies.
|
|
#
?
Dec 16, 2013 21:51
|
|
|
|
| # ? May 13, 2024 23:48 |
|
|
Coffee shops cluster because that's the Nash equilibrium for their locations (see: Hotelling's law. Your examples don't really make sense. Computer aided translation is not itself a patentable idea -- it's been described in science fiction for quite a long time. How one does it is the hard part and there are probably algorithms that would be patentable. Of course, if you do figure it out, the other option is to keep the algorithm secret rather than patent it, but I'm not sure how that serves the public interest more than disclosure and legal protection. The problem with a patent on play a FPS with a Kinect is obviousness and specificity, not subject matter. It's actually quite similar to the "patent the problem" patents I was object to earlier. You still haven't given a reason that software should be treated specially, rather than obvious patents being bad in any industry.
|
|
#
?
Dec 16, 2013 22:29
|
|