You'll note you used the word "Macrovision" about a half-dozen times. This is the other problem with digital convergence - the newer DRM schemes make Macrovision seem quite friendly indeed. I'm reading lots of recent mythtv emails from people whose firewire feeds are now being encrypted and they're now up the creek.
Pretty soon people will be selling TV-mods to add firewire outputs to them. They already make them for tuner boxes but they are pricey (and I'm not sure how they handle DRMed content).
And yet once all this fancy DRM becomes standard, how much impact do you think it will have on thepiratebay? All the mythtv users will be flocking there to download their HD shows since all the legit sites won't work...
The mentally challenged feel pain, happiness, sadness, and many other things, even if not like we do. An embryo does not. That's the difference.
I'd imagine that most animals are capable of feeling pain, and something tantamount to happiness and sadness. Have you ever seen a dog whimper from apparent dreaming during sleep?
I'm not saying that this isn't a legitimate field for debate, or that it is impossible for a line to be drawn. However, most people who feel they have some sort of rational standard for when "life" begins just haven't thought it through. Gosh, it is hard enough to define the term "planet". Most definitions of the start of life are essentially arbitrary or based on a religious viewpoint. I'm not sure I've seen one that has satisfied me as being rationally derived from biology in some way.
For the record, I believe in free markets whenever possible.
And yes, arbitrary taxes are certainly make the market less free.
And yes, the way people talk about taxing SUVs to death is bordering on imposing arbitrary taxes.
However, I would support higher taxes on fuel - but for a very economical reason - externalities.
Right now the cost of fuel is subsidized by US military spending in the middle east, environmental cleanup spending, and health care spending associated with pollution. These make fossil fuels seem artificially cheap compared to their true cost to society, so the market is dysfunctional.
The costs of all these things should be factored into a gasoline tax (with similar penalties on other fuels as appropriate). Then we should stop bashing SUV-drivers - let them spend their money how they will. The market will very efficiently select for the most efficient modes of transportation in light of their true costs to society - which will probably not be SUVs.
Environmentalists will no longer be forced to subsidize wars in the middle east. With taxes set appropriately we can also afford to send soldiers into battle with proper armor, and with half-decent benefits for those who retire or fall in the line of duty. And maybe the need to send soldiers overseas will disappear quickly if oil demand drops.
Free markets don't generally work well when there are externalities. They should rightly be turned into taxes. However, we shouldn't regulate things out of existence - we should aim only to have costs reflect the true total cost to society of any product.
I don't disagree with that. Raises an interesting question though - suppose an FAA-certified aircraft is locked down, but the non-US version isn't, could GPLv3 software then be used ? Which version of the hardware is the "recommended or principal context of use" ?
I'd imagine that they could incorporate GPLv3 code into the international version, but not the US version.
Now suppose **AA makes TIVO lock the hardware down in the US, but they do a Russian "all-of-mp4" model which is unlocked.
Is that ok with GPLv3 ?
As long as they don't use GPLv3 code in the US version.
The two versions are two different products, each with its own "recommended or principal context of use".
The whole point of the GPLv3 is to disallow Tivoization of software, so if any loopholes do arise that allow it I'm sure they'll get closed in v4...
Precisely. That is now no longer dual-key (the whole point is that no one ever has both keys) - you just restated my point that GPLv3 prohibits dual-key.
Uh, the voting commission can still ensure that no single person has both keys, thus accomplishing the goal of a dual-key system.
And even if GPLv3 was not dual-key compatible, why is that all that important? I can't really think of any applications that really require dual-key approval of software, where the software owner doesn't have both keys. So, if I write some GPLv3 code I won't lose much sleep over the fact that somebody who mistakes buzzwords for security won't be able to use my code without asking me for permission first...
To use your illustration if there is an electronic voting system I could care less whether it is dual-key, or if the software is easy or hard to tamper with. I want a printout showing how I voted, which becomes an official record of the vote. If that is done I could care less if somebody can flash the firmware since it won't make any difference except to disable the machine (and any voting location should have plenty of manual ballots on hand for any number of reasons).
And in the scenario of a bank with safety-deposit boxes the GPLv3 wouldn't get in the way at all. Suppose you have a website that requires two keys to retrieve data from it - the GPLv3 doesn't even regulate that at all. It only concerns the ability of the owner of the website to modify their own code, and possibly the ability of users to download the source for the website - not the data stored inside.
Thus GPLv3 could bar use in a legitimate field of endeavour (because of the regulatory rules in that field) that was not barred under GPLv2.
Uh, you could also word that as "Thus the FAA could bar use in a legitimate field of endeavor (because of licensing issues) that is not barred in nations with more flexible regulation." [sic]
Many regulatory schemes are bad because they regulate the how and not the what. If the goal is aircraft safety then have regulations that promote that end. They shouldn't specify how exactly the end has to be achieved - at least not to the level where they are specifying how to implement firmware.
If they want certification of every part on an aircraft then say that. Don't say that you also have to go out of your way to keep 3rd parties who are also FAA-regulated from trying to bypass the regulations.
If somebody wants to break the rules they will - and NO amount of vendor lock-in is going to stop them. All the vendor lock-in does is raise the costs of legitmate use of the equipment.
You can link GPLv2 and BSD code - as long as the resulting code is released under GPLv2.
You cannot link GPLv2 and GPLv3 code and distribute the result. GPLv3 would prohibit redistributing under anything other than GPLv3, and GPLv2 would prohibit redistributing under anything other than GPLv2. It was designed this way intentionally.
If I wrote a bunch of code for a GPLv3 project I'd be annoyed if somebody went and used it in a GPLv2 project without asking me for a license. If I wanted people copying my code without restriction I'd have made it BSD or public domain. If the author chooses GPLv3 there is probably a reason for it (maybe they don't want Tivo stealing their code withing giving back).
And what happens if the user wants to use the aircraft in Africa? Why should they need FAA-signed software when the FAA lacks jurisdiction over the end-user?
The FAA should certainly regulate the skies over America. They should even regulate manufacturing practices for aircraft flown over America. They should not require vendors to lock down their systems so that the whole world is subject to the cost of FAA regulations.
Simple - you provide the owner of the system both keys.
If it is voting, give the voting commission both keys.
Voters don't own the voting machines, therefore voters don't need the keys. And the user doesn't have to be a human entity if the software is distributed to a corporation.
In the case of aircraft I'd argue that whoever owns the aircraft should be able to do whatever they want with it software-wise. Whether or not they can legally fly it after doing so is a separate issue - and not a matter that the software vendor should be concerned with. Should aircraft maintenance panels have locks that only the vendor can open? It is up to end-users to obey the law if the FAA requires certification of flight control software. If a user wants to install software which is legal for use in flight but which the vendor hasn't blessed, then the user should be able to do so. Perhaps the user wants to fly the plane over Botswana, or maybe the user got an FAA blessing for the software despite vendor non-cooperation.
In any case, what security does the dual-key concept provide in any of these situations? You don't need both keys to rig the machine - it just makes it a lot easier.
Uh, couldn't somebody else just register a new type of aircraft - a Boeing 737 with 3rd party bolts? And then apply for airworthiness? If they did that, then the 3rd party is now an official original manufacturer (of the modified plane). Viola, no problem for people to use non-Boeing-approved parts.
I'm all for a requirement of validation of parts in safety-sensitive arenas. However, there are ways of accomplishing this without giving vendors monopoly powers over their parts.
In any case, people should still be free to modify the firmware on their planes - they just fly them if they do so. If the plane manufacturer doesn't like it they can just avoid using GPL v3 code. Just like Tivo will have to do so in the future if things go the GPL v3 route.
I'm just tired of people who complain about how a certain apps use more memory in new versions, when the programmer intentionally did that in order to improve performance or add some feature to the application.
Swapping to disk is very slow compared to caching something in memory. In older programs, it was often necessary to only have in memory what you were working on at that very moment in time.
Actually, I tend to get the best performance by keeping -Os in my CFLAGS - smaller code means more cache hits and less swapping.
Some memory/speed tradeoffs are probably worth it. However, with modern VMs this isn't always the case. If you're going to dump stuff to disk you might as well do it and read it back when needed - or MMAP the file - the OS will swap in and out for you. If you need to hold a buffer temporarily you're best off just keeping it in RAM - let the OS decide when to swap it out (you might never dump it to disk that way).
Basically, you need to keep in mind that sometimes the OS can do a better job managing memory than you can, and sometimes the reverse is true. If you put a lot of thought into your algorithm and it is tailored to what it needs to do, then it will probably beat the OS. If you just tossed something together the OS might do a better job. And of course if you use a popular library you're probably going to make good use of RAM.
The article didn't say that the company wasn't allowed to conduct 100% testing, I think it was more along the lines of not allowing them to label their meat as 100% tested, which ends up being the same thing.
I think it was worth much less the profiteers were taking advantage of a situation to maximize their profits.
If the drug is necessary to save lives, then it is "worth" quite a lot. Are you suggesting that life-saving medicines are not that valuable to society?
Your issue is that they were trying to charge what the product was "worth" - which is a different issue. Let's explore that.
First, as others have suggested, it probably didn't cost anywhere near that level - you might have been getting an exaggerated value.
Second, suppose it cost $100/day (still steep, but more in line with some rare medicines). Drugs priced at this level tend to be infrequently used (like single-shot drugs used in hospitals for obscure problems or in unusual situations). They cost the same to R&D as any other drug, or possibly more (ever try to find 5000 people with an obscure disease for a clinical trial?). If only 100,000 doses of the drug will be sold in 10 years of patent life, and the drug costs $500 million to R&D, do the math and you'll see that the doses will be expensive. The alternative is just not bothering to develop a drug that will help so few people. Usually the way these drugs actually get developed is that they were originally targeted at something else, but didn't pan out, but it was worth the small extra investment to at least try to get something back from the drug.
Most brand-name drugs that are commonly prescribed cost $80-200/month. They often treat problems that used to require surgery, so they actually save patients money compared to what they would have had to pay 10 years ago. Most drugs generate profits of a few hundred million dollars a yea. Some make only a few tens of millions, and some make a few billion, but those are rare (they just tend to make the headlines). They all cost about half a billion for R&D - which gets paid up-front before the drugs true value is known, and before all the side effects are known. Many drugs never recoup this money. The blockbusters do and more, but if you limited their profits to just a little over initial costs nobody would bother to try to find them - because most of the time you would find drugs that would barely break even. The blockbusters make up for that, but only while they're allowed to make huge profits.
If you want public R&D then get congress to fund the NIH to do the full R&D cycle for a few drugs and offer them patent-free for anyone to market. Then the market can choose between the public and private drugs, and the public can decide if the public investment is working out. You don't need to ban patents - just offer an alternative. If public medicine really works then it will be able to compete. If it doesn't, then we haven't lost all that much by trying and we'll still have the status quo.
What doesn't work is trying to magically pretend that if you limit drug prices that people will still bother to make new drugs. They'll certainly continue to market existing drugs at lower prices - those costs are sunk. However, nobody invests a half-billion dollars if they're guaranteed NOT to get a return.
And why do they spend on marketing? To make more money. Which means more money for R&D.
In any case, the big pharma companies have already kept their promise regarding spending on R&D - they spent it before the drug ever hit the market.
Sure, this year's profits go to next year's products, but this year's product cost money last year. The first drugs to hit the market were paid for by entrepeneurs who had no drug profits at all.
So, pharma would be paying for R&D even if they stopped all R&D entirely - the R&D investment was made in the past.
That's how all business works - take a risk today, expect a return tomorrow. You can't claim to be fair and deny the return after the business has already taken the risk.
In any case, if you really thing that public medicine is better why not ask your elected official to spend more on public drug development, with no patents on the resulting drugs (or defensive patents only). Don't ban private industry - just compete with it. Then everybody can look at the pros and cons of each model and see what makes the most sense.
We'll see how the "capitalists" deal with having one of their drugs undercut by researchers. I'm sure a "doctrine of equivalents" lawsuit is in the works right this instant to kill the new competition.
Nah - they'll just sit back.
It will be 10 years and hundreds of millions of dollars later before these scientists will even know if their "new drug" actually works. After spending all that money their university will force them to patent the new drug to recoup their costs, and will probably sell it to a major pharma company.
Scientists announce new drugs all the time. What they really have are molecules that have some effect in a test tube. The difference between that and a drug is about 50 refinements on the molecule and a lot of dogs and human test subjects - and those don't come cheap.
Drug companies should be obligated to allow at least one generic per name-brand drug they produce and patent. It may mean a little less advertising and Dr. smoozing money, but the rewards are more valuable than any amount of money.
Uh, drug companies would make very little money (a few million dollars per year - vs hundreds of millions per year) at all under such a system. The generic drug would be sold slightly above cost, and so the branded drug would be about the same price. And why would anybody buy the branded drug at all?
When drug R&D costs hundreds of millions of dollars per drug, why would you invest it on a product that might make a few million per year? Your breakeven time would be a century.
If you just want unpatented drugs you don't need to ban anything or regulate the industry - just ask your congressman to direct the NIH to spend money on drug R&D and release the results into the public domain. If the public pays for a drug it should be freely licensable. Right now the public only does some basic-reasearch - not full drug R&D.
Then consumers can choose between the cheap public drugs and the expensive private ones, and we can look at how much the extra public R&D is costing vs saving - then we can make an informed decision as to whether to continue the model. And as a bonus we haven't had to regulate anything and if the public model doesn't work out we haven't lost anything either.
I don't think that the FDA is concerned with any of this - the issue is with the USDA. Granted the FDA isn't a whole lot better with regard to enforcing tradition, but they at least attempt to be progressive.
How about "100% screened to reduce incidence of BSE" with a disclaimer underneath explaining the pros and cons of their approach.
Or how about no label claim at all - just brand reputation and private communications with foreign regulators (who are scientists who fully understand the ramifications of their methodology).
This isn't about truth-in-advertising. It is about setting a limit on testing to try to reduce industry-wide costs. Essentially it is quality-fixing instead of price-fixing. Similar stuff was done with airlines before prices were deregulated - there was a limit on the quality of meals served so that airlines couldn't entice customers by offering better amenities instead of lower prices.
Until recently most large java apps would not run on 64-bit linux, because of the buggy VM implementations. Sure, anything small worked fine, but try Freenet or Eclipse or anything like that and you were up the creek. And many apps would work with Sun's JRE and not with Blackdown, and vice-versa. It got to the point where I just built everything without java support.
Now things seem to have improved quite a bit, and I expect that the GPL will help them quite a bit. However, I can feel the original poster's pain.
Note - I am not a Ruby expert, but 5+3 == 5.+(3) == 4.next+2.next - literals are objects and have member functions (which they inherit from some underlying class). Operators are just shortcuts for calling functions (5+3 == 8 is the same as 5.+(3).==(8) ).
Anybody with more knowledge of Ruby feel free to chime in - I'm just starting to learn Rails in my spare time.
If an app wastes RAM then it WILL slow down the system, even if it happens to use CPU cycles slightly more efficiently.
RAM that is unused gets used by the OS for buffering and caching. So, even if the system has 50GB of RAM free, it will run faster if your application uses only 3GB rather than 4GB.
And $100/GB isn't cheap - not when CPUs sell for $200-300 for a pretty nice model. In most systems RAM tends to remain rate-limiting, and so it needs to be treated as a valuable resource and not wasted.
Password expiration is likely to result in the use of weaker/written-down passwords, and rotation schemes that diminish security while not actually limiting the duration of a breach. Should it be illegal to employ them.
If the USDA wanted a disclaimer on the packaging explaining the caveat that the testing is not reliable on young cows as used in packaged meat that would be perfectly fine. Of course, the same caveat would apply to ALL distributed meat (whether they employ random spot-testing or 100% testing).
I'd still argue that 100% tested meat is better than untested meat, even if the testing isn't 100% reliable. And if there were a market for it maybe somebody would come up with a better test.
Truth in advertising is a perfectly good application of government regulation. Prevention of testing the government feels is unnecessary is not.
I happen to believe that irradiated and genetically-modified foods are perfectly safe, but I don't oppose nations that require them to be labeled. The public might consist of idiots but they still have a right to know what they're buying. They also tend to make wrong choices on election day but I'm not about to suggest naming myself as dictator either (even though this would clearly solve the world's problems).
What they really need to do is have a standard for caching the optimized code. You run a Java app for 10 minutes, and it finally starts to really shine. Then you toss all your hard work and go back to bytecode on the next run. If they just cached the native code it would start out fast the next time, and still get faster.
Most gui-oriented apps run for a long time. But, most command-line apps don't. What was the longest grep you ever ran? Ditto for sed/awk/etc. Lots of apps get spawned and die just as fast - this is where python works great and java just doesn't.
Slashdot Mirror
You'll note you used the word "Macrovision" about a half-dozen times. This is the other problem with digital convergence - the newer DRM schemes make Macrovision seem quite friendly indeed. I'm reading lots of recent mythtv emails from people whose firewire feeds are now being encrypted and they're now up the creek.
Pretty soon people will be selling TV-mods to add firewire outputs to them. They already make them for tuner boxes but they are pricey (and I'm not sure how they handle DRMed content).
And yet once all this fancy DRM becomes standard, how much impact do you think it will have on thepiratebay? All the mythtv users will be flocking there to download their HD shows since all the legit sites won't work...
I'd imagine that most animals are capable of feeling pain, and something tantamount to happiness and sadness. Have you ever seen a dog whimper from apparent dreaming during sleep?
I'm not saying that this isn't a legitimate field for debate, or that it is impossible for a line to be drawn. However, most people who feel they have some sort of rational standard for when "life" begins just haven't thought it through. Gosh, it is hard enough to define the term "planet". Most definitions of the start of life are essentially arbitrary or based on a religious viewpoint. I'm not sure I've seen one that has satisfied me as being rationally derived from biology in some way.
For the record, I believe in free markets whenever possible.
And yes, arbitrary taxes are certainly make the market less free.
And yes, the way people talk about taxing SUVs to death is bordering on imposing arbitrary taxes.
However, I would support higher taxes on fuel - but for a very economical reason - externalities.
Right now the cost of fuel is subsidized by US military spending in the middle east, environmental cleanup spending, and health care spending associated with pollution. These make fossil fuels seem artificially cheap compared to their true cost to society, so the market is dysfunctional.
The costs of all these things should be factored into a gasoline tax (with similar penalties on other fuels as appropriate). Then we should stop bashing SUV-drivers - let them spend their money how they will. The market will very efficiently select for the most efficient modes of transportation in light of their true costs to society - which will probably not be SUVs.
Environmentalists will no longer be forced to subsidize wars in the middle east. With taxes set appropriately we can also afford to send soldiers into battle with proper armor, and with half-decent benefits for those who retire or fall in the line of duty. And maybe the need to send soldiers overseas will disappear quickly if oil demand drops.
Free markets don't generally work well when there are externalities. They should rightly be turned into taxes. However, we shouldn't regulate things out of existence - we should aim only to have costs reflect the true total cost to society of any product.
I don't disagree with that. Raises an interesting question though - suppose an FAA-certified aircraft is locked down, but the non-US version isn't, could GPLv3 software then be used ? Which version of the hardware is the "recommended or principal context of use" ?
I'd imagine that they could incorporate GPLv3 code into the international version, but not the US version.
Now suppose **AA makes TIVO lock the hardware down in the US, but they do a Russian "all-of-mp4" model which is unlocked.
Is that ok with GPLv3 ?
As long as they don't use GPLv3 code in the US version.
The two versions are two different products, each with its own "recommended or principal context of use".
The whole point of the GPLv3 is to disallow Tivoization of software, so if any loopholes do arise that allow it I'm sure they'll get closed in v4...
Precisely. That is now no longer dual-key (the whole point is that no one ever has both keys) - you just restated my point that GPLv3 prohibits dual-key.
Uh, the voting commission can still ensure that no single person has both keys, thus accomplishing the goal of a dual-key system.
And even if GPLv3 was not dual-key compatible, why is that all that important? I can't really think of any applications that really require dual-key approval of software, where the software owner doesn't have both keys. So, if I write some GPLv3 code I won't lose much sleep over the fact that somebody who mistakes buzzwords for security won't be able to use my code without asking me for permission first...
To use your illustration if there is an electronic voting system I could care less whether it is dual-key, or if the software is easy or hard to tamper with. I want a printout showing how I voted, which becomes an official record of the vote. If that is done I could care less if somebody can flash the firmware since it won't make any difference except to disable the machine (and any voting location should have plenty of manual ballots on hand for any number of reasons).
And in the scenario of a bank with safety-deposit boxes the GPLv3 wouldn't get in the way at all. Suppose you have a website that requires two keys to retrieve data from it - the GPLv3 doesn't even regulate that at all. It only concerns the ability of the owner of the website to modify their own code, and possibly the ability of users to download the source for the website - not the data stored inside.
Thus GPLv3 could bar use in a legitimate field of endeavour (because of the regulatory rules in that field) that was not barred under GPLv2.
Uh, you could also word that as "Thus the FAA could bar use in a legitimate field of endeavor (because of licensing issues) that is not barred in nations with more flexible regulation." [sic]
Many regulatory schemes are bad because they regulate the how and not the what. If the goal is aircraft safety then have regulations that promote that end. They shouldn't specify how exactly the end has to be achieved - at least not to the level where they are specifying how to implement firmware.
If they want certification of every part on an aircraft then say that. Don't say that you also have to go out of your way to keep 3rd parties who are also FAA-regulated from trying to bypass the regulations.
If somebody wants to break the rules they will - and NO amount of vendor lock-in is going to stop them. All the vendor lock-in does is raise the costs of legitmate use of the equipment.
Uh, what makes the GPL great? Being just like the BSD license? :)
You can link GPLv2 and BSD code - as long as the resulting code is released under GPLv2.
You cannot link GPLv2 and GPLv3 code and distribute the result. GPLv3 would prohibit redistributing under anything other than GPLv3, and GPLv2 would prohibit redistributing under anything other than GPLv2. It was designed this way intentionally.
If I wrote a bunch of code for a GPLv3 project I'd be annoyed if somebody went and used it in a GPLv2 project without asking me for a license. If I wanted people copying my code without restriction I'd have made it BSD or public domain. If the author chooses GPLv3 there is probably a reason for it (maybe they don't want Tivo stealing their code withing giving back).
And what happens if the user wants to use the aircraft in Africa? Why should they need FAA-signed software when the FAA lacks jurisdiction over the end-user?
The FAA should certainly regulate the skies over America. They should even regulate manufacturing practices for aircraft flown over America. They should not require vendors to lock down their systems so that the whole world is subject to the cost of FAA regulations.
Simple - you provide the owner of the system both keys.
If it is voting, give the voting commission both keys.
Voters don't own the voting machines, therefore voters don't need the keys. And the user doesn't have to be a human entity if the software is distributed to a corporation.
In the case of aircraft I'd argue that whoever owns the aircraft should be able to do whatever they want with it software-wise. Whether or not they can legally fly it after doing so is a separate issue - and not a matter that the software vendor should be concerned with. Should aircraft maintenance panels have locks that only the vendor can open? It is up to end-users to obey the law if the FAA requires certification of flight control software. If a user wants to install software which is legal for use in flight but which the vendor hasn't blessed, then the user should be able to do so. Perhaps the user wants to fly the plane over Botswana, or maybe the user got an FAA blessing for the software despite vendor non-cooperation.
In any case, what security does the dual-key concept provide in any of these situations? You don't need both keys to rig the machine - it just makes it a lot easier.
Uh, couldn't somebody else just register a new type of aircraft - a Boeing 737 with 3rd party bolts? And then apply for airworthiness? If they did that, then the 3rd party is now an official original manufacturer (of the modified plane). Viola, no problem for people to use non-Boeing-approved parts.
I'm all for a requirement of validation of parts in safety-sensitive arenas. However, there are ways of accomplishing this without giving vendors monopoly powers over their parts.
In any case, people should still be free to modify the firmware on their planes - they just fly them if they do so. If the plane manufacturer doesn't like it they can just avoid using GPL v3 code. Just like Tivo will have to do so in the future if things go the GPL v3 route.
Well, how else are you going to understand how all those tubes work! Don't you want to be qualified to work on the internet?
I'm just tired of people who complain about how a certain apps use more memory in new versions, when the programmer intentionally did that in order to improve performance or add some feature to the application.
Swapping to disk is very slow compared to caching something in memory. In older programs, it was often necessary to only have in memory what you were working on at that very moment in time.
Actually, I tend to get the best performance by keeping -Os in my CFLAGS - smaller code means more cache hits and less swapping.
Some memory/speed tradeoffs are probably worth it. However, with modern VMs this isn't always the case. If you're going to dump stuff to disk you might as well do it and read it back when needed - or MMAP the file - the OS will swap in and out for you. If you need to hold a buffer temporarily you're best off just keeping it in RAM - let the OS decide when to swap it out (you might never dump it to disk that way).
Basically, you need to keep in mind that sometimes the OS can do a better job managing memory than you can, and sometimes the reverse is true. If you put a lot of thought into your algorithm and it is tailored to what it needs to do, then it will probably beat the OS. If you just tossed something together the OS might do a better job. And of course if you use a popular library you're probably going to make good use of RAM.
The article didn't say that the company wasn't allowed to conduct 100% testing, I think it was more along the lines of not allowing them to label their meat as 100% tested, which ends up being the same thing.
Ok, how about this?
My thoughts is that this could be handled a different way, but from what I've seen articles frequently get facts wrong.
Sure, but a little googling shows that there is more to this than one article.
I think it was worth much less the profiteers were taking advantage of a situation to maximize their profits.
If the drug is necessary to save lives, then it is "worth" quite a lot. Are you suggesting that life-saving medicines are not that valuable to society?
Your issue is that they were trying to charge what the product was "worth" - which is a different issue. Let's explore that.
First, as others have suggested, it probably didn't cost anywhere near that level - you might have been getting an exaggerated value.
Second, suppose it cost $100/day (still steep, but more in line with some rare medicines). Drugs priced at this level tend to be infrequently used (like single-shot drugs used in hospitals for obscure problems or in unusual situations). They cost the same to R&D as any other drug, or possibly more (ever try to find 5000 people with an obscure disease for a clinical trial?). If only 100,000 doses of the drug will be sold in 10 years of patent life, and the drug costs $500 million to R&D, do the math and you'll see that the doses will be expensive. The alternative is just not bothering to develop a drug that will help so few people. Usually the way these drugs actually get developed is that they were originally targeted at something else, but didn't pan out, but it was worth the small extra investment to at least try to get something back from the drug.
Most brand-name drugs that are commonly prescribed cost $80-200/month. They often treat problems that used to require surgery, so they actually save patients money compared to what they would have had to pay 10 years ago. Most drugs generate profits of a few hundred million dollars a yea. Some make only a few tens of millions, and some make a few billion, but those are rare (they just tend to make the headlines). They all cost about half a billion for R&D - which gets paid up-front before the drugs true value is known, and before all the side effects are known. Many drugs never recoup this money. The blockbusters do and more, but if you limited their profits to just a little over initial costs nobody would bother to try to find them - because most of the time you would find drugs that would barely break even. The blockbusters make up for that, but only while they're allowed to make huge profits.
If you want public R&D then get congress to fund the NIH to do the full R&D cycle for a few drugs and offer them patent-free for anyone to market. Then the market can choose between the public and private drugs, and the public can decide if the public investment is working out. You don't need to ban patents - just offer an alternative. If public medicine really works then it will be able to compete. If it doesn't, then we haven't lost all that much by trying and we'll still have the status quo.
What doesn't work is trying to magically pretend that if you limit drug prices that people will still bother to make new drugs. They'll certainly continue to market existing drugs at lower prices - those costs are sunk. However, nobody invests a half-billion dollars if they're guaranteed NOT to get a return.
And why do they spend on marketing? To make more money. Which means more money for R&D.
In any case, the big pharma companies have already kept their promise regarding spending on R&D - they spent it before the drug ever hit the market.
Sure, this year's profits go to next year's products, but this year's product cost money last year. The first drugs to hit the market were paid for by entrepeneurs who had no drug profits at all.
So, pharma would be paying for R&D even if they stopped all R&D entirely - the R&D investment was made in the past.
That's how all business works - take a risk today, expect a return tomorrow. You can't claim to be fair and deny the return after the business has already taken the risk.
In any case, if you really thing that public medicine is better why not ask your elected official to spend more on public drug development, with no patents on the resulting drugs (or defensive patents only). Don't ban private industry - just compete with it. Then everybody can look at the pros and cons of each model and see what makes the most sense.
We'll see how the "capitalists" deal with having one of their drugs undercut by researchers. I'm sure a "doctrine of equivalents" lawsuit is in the works right this instant to kill the new competition.
Nah - they'll just sit back.
It will be 10 years and hundreds of millions of dollars later before these scientists will even know if their "new drug" actually works. After spending all that money their university will force them to patent the new drug to recoup their costs, and will probably sell it to a major pharma company.
Scientists announce new drugs all the time. What they really have are molecules that have some effect in a test tube. The difference between that and a drug is about 50 refinements on the molecule and a lot of dogs and human test subjects - and those don't come cheap.
Drug companies should be obligated to allow at least one generic per name-brand drug they produce and patent. It may mean a little less advertising and Dr. smoozing money, but the rewards are more valuable than any amount of money.
Uh, drug companies would make very little money (a few million dollars per year - vs hundreds of millions per year) at all under such a system. The generic drug would be sold slightly above cost, and so the branded drug would be about the same price. And why would anybody buy the branded drug at all?
When drug R&D costs hundreds of millions of dollars per drug, why would you invest it on a product that might make a few million per year? Your breakeven time would be a century.
If you just want unpatented drugs you don't need to ban anything or regulate the industry - just ask your congressman to direct the NIH to spend money on drug R&D and release the results into the public domain. If the public pays for a drug it should be freely licensable. Right now the public only does some basic-reasearch - not full drug R&D.
Then consumers can choose between the cheap public drugs and the expensive private ones, and we can look at how much the extra public R&D is costing vs saving - then we can make an informed decision as to whether to continue the model. And as a bonus we haven't had to regulate anything and if the public model doesn't work out we haven't lost anything either.
I don't think that the FDA is concerned with any of this - the issue is with the USDA. Granted the FDA isn't a whole lot better with regard to enforcing tradition, but they at least attempt to be progressive.
How about "100% screened to reduce incidence of BSE" with a disclaimer underneath explaining the pros and cons of their approach.
Or how about no label claim at all - just brand reputation and private communications with foreign regulators (who are scientists who fully understand the ramifications of their methodology).
This isn't about truth-in-advertising. It is about setting a limit on testing to try to reduce industry-wide costs. Essentially it is quality-fixing instead of price-fixing. Similar stuff was done with airlines before prices were deregulated - there was a limit on the quality of meals served so that airlines couldn't entice customers by offering better amenities instead of lower prices.
Why would that law ever be changed in favor of consumers? I can't imagine a scenario where that would fly.
:)
I dunno - maybe a scenario where consumers wrote the laws?
Until recently most large java apps would not run on 64-bit linux, because of the buggy VM implementations. Sure, anything small worked fine, but try Freenet or Eclipse or anything like that and you were up the creek. And many apps would work with Sun's JRE and not with Blackdown, and vice-versa. It got to the point where I just built everything without java support.
Now things seem to have improved quite a bit, and I expect that the GPL will help them quite a bit. However, I can feel the original poster's pain.
In Ruby - EVERYTHING is an object.
Note - I am not a Ruby expert, but 5+3 == 5.+(3) == 4.next+2.next - literals are objects and have member functions (which they inherit from some underlying class). Operators are just shortcuts for calling functions (5+3 == 8 is the same as 5.+(3).==(8) ).
Anybody with more knowledge of Ruby feel free to chime in - I'm just starting to learn Rails in my spare time.
If an app wastes RAM then it WILL slow down the system, even if it happens to use CPU cycles slightly more efficiently.
RAM that is unused gets used by the OS for buffering and caching. So, even if the system has 50GB of RAM free, it will run faster if your application uses only 3GB rather than 4GB.
And $100/GB isn't cheap - not when CPUs sell for $200-300 for a pretty nice model. In most systems RAM tends to remain rate-limiting, and so it needs to be treated as a valuable resource and not wasted.
Password expiration is likely to result in the use of weaker/written-down passwords, and rotation schemes that diminish security while not actually limiting the duration of a breach. Should it be illegal to employ them.
If the USDA wanted a disclaimer on the packaging explaining the caveat that the testing is not reliable on young cows as used in packaged meat that would be perfectly fine. Of course, the same caveat would apply to ALL distributed meat (whether they employ random spot-testing or 100% testing).
I'd still argue that 100% tested meat is better than untested meat, even if the testing isn't 100% reliable. And if there were a market for it maybe somebody would come up with a better test.
Truth in advertising is a perfectly good application of government regulation. Prevention of testing the government feels is unnecessary is not.
I happen to believe that irradiated and genetically-modified foods are perfectly safe, but I don't oppose nations that require them to be labeled. The public might consist of idiots but they still have a right to know what they're buying. They also tend to make wrong choices on election day but I'm not about to suggest naming myself as dictator either (even though this would clearly solve the world's problems).
What they really need to do is have a standard for caching the optimized code. You run a Java app for 10 minutes, and it finally starts to really shine. Then you toss all your hard work and go back to bytecode on the next run. If they just cached the native code it would start out fast the next time, and still get faster.
Most gui-oriented apps run for a long time. But, most command-line apps don't. What was the longest grep you ever ran? Ditto for sed/awk/etc. Lots of apps get spawned and die just as fast - this is where python works great and java just doesn't.