I don't understand. When SATA was first announced as a competitor to Firewire, all the anything-but-apple proponents said the reason it was so wonderful and necessary was that it would be completely compatible with ATA, no need for new drivers, isn't that wonderful. The SATA standard itself says that a SATA card must emulate an ATA device (master only, optionally emulate master/slave). So how can it be that these devices don't just work right out of the box, with full support for SATA required for higher speed, maybe higher capacity, but otherwise just plain working?
The whole interconnect wars is just plain stupid. It was stupid back when it was Apple and SCSI vs. IBM PC and various standards, eventually IDE and EIDE, then ATAPI (which is just SCSI over IDE). Now we have Firewire (which is sort of SCSI over serial), SATA (which is ATA over serial, so with ATAPI that makes it... SCSI over serial), and of course Serial Attached SCSI (which is SCSI over a serial link). WTF??? Let's not even talk about USB, which started off as a slow-speed bus but is now trying to compete with all of the above as well. So now we get hard drives which do USB 2.0 and Firewire (400? 800?) as external drives, and along comes, TADA, EXTERNAL SATA.
So the modern computer will have Firewire 800, USB 2, serial attached SCSI (which can handle SATA as well), Gigabit Ethernet - how many standards do we need for transmitting bits!??
"That's the nice thing about standards - there are so many to choose from, and if you don't like any of those, you can always create your own new standard!"
But playing games inside is higher priority than doing legitimate research outside? Remember, this is wireless access. Inside, I could be doing just about anything on MY OWN computer, even sitting such that no one else can see the screen, maybe sitting in a corner, or on a couch. I really don't think you'll have librarians going around snooping to see whether people are doing legitimate research or just reading/.
a) the library was closed at the time, so no patrons inside were being denied bandwidth; b) he is a library patron, whether he was inside or outside the library. Him walking inside isn't going to magically make him use less bandwidth. You can't even use the argument that a coffee shop might use, that they're providing the service for their customers so that they get more revenue - the library isn't going to sell more merchandise if he goes indoors to use their network.
Do I think the library has the right to say whether he can use it outdoors? I reluctantly think yes, but I could be convinced otherwise. I'm mostly troubled by the presence of a completely unprotected signal - to me, that is an invitation to use it, and impossible to distinguish from an open network that actually DOES want you to use it. Do I think a taxpayer-funded library SHOULD have such restrictions on their network? No, unless it can actually be shown to be causing problems that can't be solved any other way.
As I read it a few days ago, the library didn't say you couldn't use it outside, they said that the wireless network would only be available during specific hours. Now, whether this was a policy thing ("we won't let anyone use the network...") or merely a warning that they normally wouldn't keep the network running outside of library hours, wasn't clear. It could well mean that you're free to use it if its running.
One other thing to note - in the original incident, the library was CLOSED, which is why he was accessing it from outside. He also said there were a couple of other access points wide open at that same point, and that the library had a SECURED wireless network as well (presumably for staff use, etc).
The really outrageous part was the police officer telling him he to move along, he wasn't allowed to use a computer on a public bench, whether or not he was accessing a wireless network, with or without permission. Only slightly less outrageous was the bit about how they had all been told about this Federal Law that said he was Stealing Signals since he was outside.
Where does everyone get the idea that the DMCA outlaws decrypting things? DMCA only protects copyrighted works from unauthorized access, and it isn't even necessarily encryption that has to be broken for that. There's plenty wrong with various aspects of the DMCA, but outlawing decryption of a signal is not one of them. There are other laws for that...
If utility of using GPL software outweighs (perceived) cost of contributing changes back, then you'll use GPL. Using BSD, but not contributing back to BSD, doesn't help BSD in the slightest. Switching from GPL to BSD will gain you nothing - those who weren't using it because they didn't want to contribute don't help you at all, and you lose support from those who would contribute under GPL protection but won't under BSD (because competitors can get unfair advantage, since they can take but not give back to you).
Game theory shows a good strategy in playing "Iterated Prisoner's Dilemma" is "tit-for-tat". It seems to me that GPL is an implementation of that strategy, rational companies (if they exist) would use GPL and collectively outperform those companies that refuse to participate.
Without basically altruistic input, BSD-licensed code will have a tendency to stagnate, and GPL-licensed code will have a tendency to increase exponentially, as only a portion of those who use BSD will contribute back to it, and the usage of each will be based on the utility of using it, which will increase as more contributions are made. Altruistic input changes that, of course, but is present in both license models. Strong enough input for BSD will keep it alive regardless, but will become increasingly more difficult to stand against GPL. Ironically, for a company that wants the BSD model to succeed and the GPL model to fail, the only way to do so is by contributing to BSD on a continuing basis, which seems to negate the reasons for not wanting the GPL to succeed in the first place.
Except that if it is publicly disclosed by yourself or anyone else more than one year before you FILE, it doesn't matter if you invented it first, you can't patent it. Disclosure can consist of USING the patent publicly, even if you don't disclose how you're doing it.
Although an argument can be made that su (and sudo) trump this patent, an even better case can be made for any type of server which allows you to do things your signon isn't allowed to do directly. Examples include an X-server (access to video hardware, giving conditional access by way of a magic cookie), a sound daemon, an NFS server, syslogd. Any network daemon that uses the ident protocol to conditionally give access, or rsh or ssh to a different signon than your own also seems to completely cover some of the claims. I didn't look closely at all of the dependent claims, but of the ones I couldn't immediately think of a prior example, the differences from the independent claim (all of which I do think are covered by prior art) seem trivially obvious. Also, note that half of the claims are for "media containing a computer program that..." does the same thing as the other half of the claims, so those don't even need to be looked at (putting a program on a computer-readable media for the purpose of running it on a computer is the very epitome of obviousness). Half of the remaining claims are for a "network appliance", where the web-server equivalent is running executable requests which run as a non-privileged process, using a privileged process to conditionally implement privileged requests. These have the same set of variant dependent claims as the remaining claims.
As far back as at latest 1980, the PLATO system had a facility for running programs under NOS which ran as non-privileged users, which could make requests of a privileged program to get access to files, based on the user ID (including group membership, one of the dependent claims!), looking up the access to the file in fields in the file itself. Later proposals, never actually implemented, also would have allowed such access to be mediated by an access list associated with the file (not implemented doesn't necessarily mean it can't be used as prior art - it was well discussed, obvious how to do it, but judged not worth the effort to implement given that the facility itself wasn't used by non-privileged people anyway, and there was an easy override to give the job system privileges when needed to bypass file security).
Another example from PLATO includes submitting requests for files to be transferred from one system to another - such requests were mediated by an access list on both the sending and receiving system; said access list included four levels of membership (hierarchically: system, account, group, individual), of which any of the lower levels could be a wildcard (any, or user type). This was prior to 1990, and would seem to cover many of the claims all by itself.
If you feel that strongly that it is unethical, maybe you should use a modified BSD license that doesn't allow re-licensing under more restrictive terms! So remind me again, why do you use the BSD license if you're going to complain when people use the one difference it has from the GPL? It seems hypocritical to advocate "freedom" in the BSD sense, then object when someone uses that freedom (whether to GPL modifications, or to take it completely closed).
Of course, the great thing about hypocrisy is that once you've accepted it in yourself, you can still criticize it in others...
No, you would have to show that the two acts (providing a copy and receiving a copy) were dependent. Defining "financial gain" as including receiving a copy of something of purported value is one thing, but it doesn't change the basic equation of "exchange". There has to be a tie between the two things.
Let's say your interpretation is correct. It leaves no limits, then. Your employer paying you could be every much "financial gain" as receiving a copy of something over the Internet, or getting a free AOL disk in the mail for that matter. Note that the received copy, as referenced in the law, doesn't necessarily have to be an unauthorized copy. If I offer you a copy of a program I wrote in exchange for a copy of a song you ripped, it would still constitute "financial gain" as defined in that statute. You sending out an unauthorized copy of something doesn't then make it illegal for you to download a GPLed program, for example, even though you provided and received over the Internet.
Requiring you to share does not even suggest commiting infringment, much less require it.
Ah, the person who says "share or be banned" is perfectly happy if I share items that are free to be redistributed, and nothing else? They're not suggesting that if you don't provide access to a multitude of items you're not authorized to redistribute, they will take away something of value to you? What exactly do you think "share or be banned" means, then?
How about making it illegal for anyone to be required to show their driver's license unless they are being charged with violating a law? Add in prohibiting individuals or businesses from using a driver's license for any purpose other than determining if the person is qualified to drive a vehicle (e.g. when renting a car, or employing someone who will be driving as part of their job). Same thing should go for Social Security Number - it should ONLY be used for administering the Social Security program. Not for collecting taxes, not for credit checks.
Prohibit the suspension or revocation of a driver's license for any reason other than not being able to drive safely (which includes being convicted of drunk driving).
Driving SHOULD be a right, not a privilege. I don't know where the "privilege, not a right" came from, I've never seen a justification for it.
Sending and then receiving should not be considered "financial gain" under that definition UNLESS sending your file was a pre-condition for being able to receive anything. However, a "share or be banned" explicitly sets such a pre-condition. In addition, it is attempting to force you to commit a copyright violation, which is incitement to infringe.
That's OK, you wouldn't be able to properly pronounce many of the cities in Illinois. For example, Cairo - is pronounced KAY-row. And I wouldn't expect you to know where it is, either.
When my sister moved to London from Illinois, we figured out that the UK is remarkably similar in size and shape to Illinois. I guess Scotland would correspond to the Chicago suburbs, since those never really end anyway.
Gosh, you're right. If you manage to get 1000 user IDs and passwords for each person on the planet, you'd only need 10,000 clusters of 18,000 processors, each of which is calculating 1 hash per nanosecond, for 10 years, to get an even chance of a match. Remember, each of those 10,000 clusters needs to have enough bandwidth to transfer 350,000 CD's worth of data per second.
Brute force is not the answer when dealing with difficulty on the order of 2^128 or above.
If now is the time to migrate, then it was actually time to migrate several years ago, to avoid being forced to migrate now. But, of course, that means it was time to migrate several years before that. Etc. SImilar reasoning says you should migrate off of whatever you're using immediately, because at some time in the future you'll have to migrate off of it. Everyone using x86 machines now should switch to using Itanium immediately.
Being able to not migrate off of something that is working for you now is very valuable. It means you don't lock yourself into something that may turn out to be a bad choice, if only you had waited a little longer. For instance, migrating to Itanium right now would be foolish. Perhaps it will turn out to be a winner. Perhaps it will turn out to be a loser. Much better to examine the situation over the next several years to determine the answer. Perhaps one of the licensees of the Alpha processor will decide to continue developing and producing them. Maybe the big winner in in 5 years will be the PPC-64.
Remember, whatever processor you end up with (either this new one, or continue with your old Alpha CPU if it still serves you well enough), it will basically last until it would have been time to switch to another more powerful system anyway. At that time, you'd have had to make a decision anyway on whether to stick with the same architecture or not. It's not like your current machines are all of a sudden going to stop working.
It is, of course, prudent to start PLANNING for a migration path now, since your simple option (just switch to a faster Alpha) is not going to be very useful for much longer, and repair and maintenance will become more and more expensive as time goes by.
You could easily give people a limited monopoly without requiring them to disclose their methods. The whole point of a patent is that, in exchange for you making it public, society will give you a monopoly on it for a limited time, after which it becomes public domain (and they no longer have to get your permission in order to use it). If patents have "nothing to do with sharing your ideas", then why are you required to disclose them in order to get a patent?
Patents are the opposite of trade secrets. With a trade secret, you take great care not to disclose your methods. In exchange, you get nothing, even if you "put in a lot of effort to invent something new to benefit society." You have to enforce your own monopoly with trade secrets - no one else can do the same thing because they don't know how and you won't tell them.
Patents do help stimulate invention, by providing a method to make a profit off of the risk involved in the process of inventing something new, even for those inventions that are easily apparent how they work. However, you won't get a patent if you don't disclose how it works, easily apparent or not. And plenty of inventions are disclosed by being patented, even if they'd probably be able to be kept as trade secrets - for one thing, it is a hedge against someone independently re-discovering it, for another you can license it out to others without having to be paranoid about someone leaking it. The essential element of a patent, though, is still disclosure, and that is the primary way it fulfills the Constitutional mandate to "promote the progress of Science", not by simply granting a monopoly to entice people to invent new things.
Yes, many things are shared openly - open source, scientific papers. It's a good part of an argument that patents may not be necessary to get people to share their ideas. However, that doesn't change the fact that a key element of patents is that they require you to disclose your methods, so that everyone can use them (as in, share!) after the patent period has expired.
Why would Google care about whether speculative investors buy into the IPO? They'd rather the person who might have bought from the speculative investor buy directly from them. That way Google gets all the money, instead of the speculator skimming some out of the middle.
I don't think you've thought about how ridiculously large 2^128 is. Go back and read what I said. The entire computational capability of the planet, running for the next 10 years, wouldn't even make a noticeable dent in a 2^128 sized space. Those 18,000-processor machines, one for each person, each processor doing femto-second hashes, would take 100 years to do 2^128. How much of a chunk do you think today's multi-nanosecond-per-hash processors would take, even if you had 18,000 of them for each person in the world, and ran them for 10 years? 0.000002 PERCENT, and that's giving a generous 5 nanoseconds per hash. How many 18,000-processor clusters do you think the government actually has access to? Less than 6 billion, I'll bet.
Note also that you're only talking about 16-bit chunks. It doesn't help you with any other sized chunks, of which there are a LOT more than 2^128.
Or, put another way, guessing at the key is identical to guessing at the decrypted message. Just guess what the message might be. You don't even have to intercept the encrypted message to do this. Just guess! "I wonder if it could be... aha, proof that Iraq has weapons of mass destruction! Wow, that's so unlikely that I randomly guessed that exact message that it must be right. It even has he word "misunderestimate" in it, it must be real!" Maybe the encrypted message is a little short? Guess that they first compressed it with gzip, then added some zero padding (or you can make a guess what the padding is, too), and a signed MD5 signature . Guess what? It's so unlikely that the message just happens to be what you guessed, in gzip format and all, with the exact same padding, and look at that signnature you guessed, that it must be the message. Wow, they managed to encrypt BOTH messages at the same time, what an advanced code!
In storage terms, every person on the Earth today would have to have 56 trillion 16-petabyte storage units to hold that many MD5 hashes. One of those disk drives would hold around 3000 years worth of uncompressed CD-quality stereo music (or only around 450 years of DVD-quality compressed video).
In computational terms, every person would have to have an 18,000-processor machine, each processor doing one hash every femto-second, and it would still take 100 years. The bandwidth required to store the hashes for just one of those machines would be able to transfer around 350 billion CDs per second.
The most ridiculous part is, I could be off by a whole bunch of orders of magnitude and it wouldn't make a difference in how totally off-scale those numbers are.
One problem with that is there isn't any good point that doesn't give a significant response from the blue or the yellow-green cones (not red - there are no red sensors, red is detected by the somewhat higher response of the yellow-green compared to green). Two possibilities for additional colors would be at the ends of sensitivity for green (both ends) and blue (towards the green), and at the points of equal sensitivity (e.g. approximately cyan).
No, even if perfectly calibrated to an individual eye, RGB can't perfectly reproduce real colors, even theoretically. For example, RGB can't create the same response in the eye as indigo or violet light does. Cyan and yellow light can only be approximated by RGB, due to the overlapping of the response curves of the three types of cones.
Slashdot Mirror
I don't understand. When SATA was first announced as a competitor to Firewire, all the anything-but-apple proponents said the reason it was so wonderful and necessary was that it would be completely compatible with ATA, no need for new drivers, isn't that wonderful. The SATA standard itself says that a SATA card must emulate an ATA device (master only, optionally emulate master/slave). So how can it be that these devices don't just work right out of the box, with full support for SATA required for higher speed, maybe higher capacity, but otherwise just plain working?
The whole interconnect wars is just plain stupid. It was stupid back when it was Apple and SCSI vs. IBM PC and various standards, eventually IDE and EIDE, then ATAPI (which is just SCSI over IDE). Now we have Firewire (which is sort of SCSI over serial), SATA (which is ATA over serial, so with ATAPI that makes it ... SCSI over serial), and of course Serial Attached SCSI (which is SCSI over a serial link). WTF??? Let's not even talk about USB, which started off as a slow-speed bus but is now trying to compete with all of the above as well. So now we get hard drives which do USB 2.0 and Firewire (400? 800?) as external drives, and along comes, TADA, EXTERNAL SATA.
So the modern computer will have Firewire 800, USB 2, serial attached SCSI (which can handle SATA as well), Gigabit Ethernet - how many standards do we need for transmitting bits!??
"That's the nice thing about standards - there are so many to choose from, and if you don't like any of those, you can always create your own new standard!"
But playing games inside is higher priority than doing legitimate research outside? Remember, this is wireless access. Inside, I could be doing just about anything on MY OWN computer, even sitting such that no one else can see the screen, maybe sitting in a corner, or on a couch. I really don't think you'll have librarians going around snooping to see whether people are doing legitimate research or just reading /.
a) the library was closed at the time, so no patrons inside were being denied bandwidth; b) he is a library patron, whether he was inside or outside the library. Him walking inside isn't going to magically make him use less bandwidth. You can't even use the argument that a coffee shop might use, that they're providing the service for their customers so that they get more revenue - the library isn't going to sell more merchandise if he goes indoors to use their network.
Do I think the library has the right to say whether he can use it outdoors? I reluctantly think yes, but I could be convinced otherwise. I'm mostly troubled by the presence of a completely unprotected signal - to me, that is an invitation to use it, and impossible to distinguish from an open network that actually DOES want you to use it. Do I think a taxpayer-funded library SHOULD have such restrictions on their network? No, unless it can actually be shown to be causing problems that can't be solved any other way.
As I read it a few days ago, the library didn't say you couldn't use it outside, they said that the wireless network would only be available during specific hours. Now, whether this was a policy thing ("we won't let anyone use the network ...") or merely a warning that they normally wouldn't keep the network running outside of library hours, wasn't clear. It could well mean that you're free to use it if its running.
One other thing to note - in the original incident, the library was CLOSED, which is why he was accessing it from outside. He also said there were a couple of other access points wide open at that same point, and that the library had a SECURED wireless network as well (presumably for staff use, etc).
The really outrageous part was the police officer telling him he to move along, he wasn't allowed to use a computer on a public bench, whether or not he was accessing a wireless network, with or without permission. Only slightly less outrageous was the bit about how they had all been told about this Federal Law that said he was Stealing Signals since he was outside.
Where does everyone get the idea that the DMCA outlaws decrypting things? DMCA only protects copyrighted works from unauthorized access, and it isn't even necessarily encryption that has to be broken for that. There's plenty wrong with various aspects of the DMCA, but outlawing decryption of a signal is not one of them. There are other laws for that...
If utility of using GPL software outweighs (perceived) cost of contributing changes back, then you'll use GPL. Using BSD, but not contributing back to BSD, doesn't help BSD in the slightest. Switching from GPL to BSD will gain you nothing - those who weren't using it because they didn't want to contribute don't help you at all, and you lose support from those who would contribute under GPL protection but won't under BSD (because competitors can get unfair advantage, since they can take but not give back to you).
Game theory shows a good strategy in playing "Iterated Prisoner's Dilemma" is "tit-for-tat". It seems to me that GPL is an implementation of that strategy, rational companies (if they exist) would use GPL and collectively outperform those companies that refuse to participate.
Without basically altruistic input, BSD-licensed code will have a tendency to stagnate, and GPL-licensed code will have a tendency to increase exponentially, as only a portion of those who use BSD will contribute back to it, and the usage of each will be based on the utility of using it, which will increase as more contributions are made. Altruistic input changes that, of course, but is present in both license models. Strong enough input for BSD will keep it alive regardless, but will become increasingly more difficult to stand against GPL. Ironically, for a company that wants the BSD model to succeed and the GPL model to fail, the only way to do so is by contributing to BSD on a continuing basis, which seems to negate the reasons for not wanting the GPL to succeed in the first place.
Except that if it is publicly disclosed by yourself or anyone else more than one year before you FILE, it doesn't matter if you invented it first, you can't patent it. Disclosure can consist of USING the patent publicly, even if you don't disclose how you're doing it.
Although an argument can be made that su (and sudo) trump this patent, an even better case can be made for any type of server which allows you to do things your signon isn't allowed to do directly. Examples include an X-server (access to video hardware, giving conditional access by way of a magic cookie), a sound daemon, an NFS server, syslogd. Any network daemon that uses the ident protocol to conditionally give access, or rsh or ssh to a different signon than your own also seems to completely cover some of the claims. I didn't look closely at all of the dependent claims, but of the ones I couldn't immediately think of a prior example, the differences from the independent claim (all of which I do think are covered by prior art) seem trivially obvious. Also, note that half of the claims are for "media containing a computer program that ..." does the same thing as the other half of the claims, so those don't even need to be looked at (putting a program on a computer-readable media for the purpose of running it on a computer is the very epitome of obviousness). Half of the remaining claims are for a "network appliance", where the web-server equivalent is running executable requests which run as a non-privileged process, using a privileged process to conditionally implement privileged requests. These have the same set of variant dependent claims as the remaining claims.
As far back as at latest 1980, the PLATO system had a facility for running programs under NOS which ran as non-privileged users, which could make requests of a privileged program to get access to files, based on the user ID (including group membership, one of the dependent claims!), looking up the access to the file in fields in the file itself. Later proposals, never actually implemented, also would have allowed such access to be mediated by an access list associated with the file (not implemented doesn't necessarily mean it can't be used as prior art - it was well discussed, obvious how to do it, but judged not worth the effort to implement given that the facility itself wasn't used by non-privileged people anyway, and there was an easy override to give the job system privileges when needed to bypass file security).
Another example from PLATO includes submitting requests for files to be transferred from one system to another - such requests were mediated by an access list on both the sending and receiving system; said access list included four levels of membership (hierarchically: system, account, group, individual), of which any of the lower levels could be a wildcard (any, or user type). This was prior to 1990, and would seem to cover many of the claims all by itself.
Special Olympics is explicitly authorized to use the word "Olympics".
If you feel that strongly that it is unethical, maybe you should use a modified BSD license that doesn't allow re-licensing under more restrictive terms! So remind me again, why do you use the BSD license if you're going to complain when people use the one difference it has from the GPL? It seems hypocritical to advocate "freedom" in the BSD sense, then object when someone uses that freedom (whether to GPL modifications, or to take it completely closed).
Of course, the great thing about hypocrisy is that once you've accepted it in yourself, you can still criticize it in others...
No, you would have to show that the two acts (providing a copy and receiving a copy) were dependent. Defining "financial gain" as including receiving a copy of something of purported value is one thing, but it doesn't change the basic equation of "exchange". There has to be a tie between the two things.
Let's say your interpretation is correct. It leaves no limits, then. Your employer paying you could be every much "financial gain" as receiving a copy of something over the Internet, or getting a free AOL disk in the mail for that matter. Note that the received copy, as referenced in the law, doesn't necessarily have to be an unauthorized copy. If I offer you a copy of a program I wrote in exchange for a copy of a song you ripped, it would still constitute "financial gain" as defined in that statute. You sending out an unauthorized copy of something doesn't then make it illegal for you to download a GPLed program, for example, even though you provided and received over the Internet.
Ah, the person who says "share or be banned" is perfectly happy if I share items that are free to be redistributed, and nothing else? They're not suggesting that if you don't provide access to a multitude of items you're not authorized to redistribute, they will take away something of value to you? What exactly do you think "share or be banned" means, then?
How about making it illegal for anyone to be required to show their driver's license unless they are being charged with violating a law? Add in prohibiting individuals or businesses from using a driver's license for any purpose other than determining if the person is qualified to drive a vehicle (e.g. when renting a car, or employing someone who will be driving as part of their job). Same thing should go for Social Security Number - it should ONLY be used for administering the Social Security program. Not for collecting taxes, not for credit checks.
Prohibit the suspension or revocation of a driver's license for any reason other than not being able to drive safely (which includes being convicted of drunk driving).
Driving SHOULD be a right, not a privilege. I don't know where the "privilege, not a right" came from, I've never seen a justification for it.
Sending and then receiving should not be considered "financial gain" under that definition UNLESS sending your file was a pre-condition for being able to receive anything. However, a "share or be banned" explicitly sets such a pre-condition. In addition, it is attempting to force you to commit a copyright violation, which is incitement to infringe.
That's OK, you wouldn't be able to properly pronounce many of the cities in Illinois. For example, Cairo - is pronounced KAY-row. And I wouldn't expect you to know where it is, either.
At least you don't live in New Mexico ("I'm sorry, we can't ship outside the United States")...
When my sister moved to London from Illinois, we figured out that the UK is remarkably similar in size and shape to Illinois. I guess Scotland would correspond to the Chicago suburbs, since those never really end anyway.
Gosh, you're right. If you manage to get 1000 user IDs and passwords for each person on the planet, you'd only need 10,000 clusters of 18,000 processors, each of which is calculating 1 hash per nanosecond, for 10 years, to get an even chance of a match. Remember, each of those 10,000 clusters needs to have enough bandwidth to transfer 350,000 CD's worth of data per second.
Brute force is not the answer when dealing with difficulty on the order of 2^128 or above.
If now is the time to migrate, then it was actually time to migrate several years ago, to avoid being forced to migrate now. But, of course, that means it was time to migrate several years before that. Etc. SImilar reasoning says you should migrate off of whatever you're using immediately, because at some time in the future you'll have to migrate off of it. Everyone using x86 machines now should switch to using Itanium immediately.
Being able to not migrate off of something that is working for you now is very valuable. It means you don't lock yourself into something that may turn out to be a bad choice, if only you had waited a little longer. For instance, migrating to Itanium right now would be foolish. Perhaps it will turn out to be a winner. Perhaps it will turn out to be a loser. Much better to examine the situation over the next several years to determine the answer. Perhaps one of the licensees of the Alpha processor will decide to continue developing and producing them. Maybe the big winner in in 5 years will be the PPC-64.
Remember, whatever processor you end up with (either this new one, or continue with your old Alpha CPU if it still serves you well enough), it will basically last until it would have been time to switch to another more powerful system anyway. At that time, you'd have had to make a decision anyway on whether to stick with the same architecture or not. It's not like your current machines are all of a sudden going to stop working.
It is, of course, prudent to start PLANNING for a migration path now, since your simple option (just switch to a faster Alpha) is not going to be very useful for much longer, and repair and maintenance will become more and more expensive as time goes by.
You could easily give people a limited monopoly without requiring them to disclose their methods. The whole point of a patent is that, in exchange for you making it public, society will give you a monopoly on it for a limited time, after which it becomes public domain (and they no longer have to get your permission in order to use it). If patents have "nothing to do with sharing your ideas", then why are you required to disclose them in order to get a patent?
Patents are the opposite of trade secrets. With a trade secret, you take great care not to disclose your methods. In exchange, you get nothing, even if you "put in a lot of effort to invent something new to benefit society." You have to enforce your own monopoly with trade secrets - no one else can do the same thing because they don't know how and you won't tell them.
Patents do help stimulate invention, by providing a method to make a profit off of the risk involved in the process of inventing something new, even for those inventions that are easily apparent how they work. However, you won't get a patent if you don't disclose how it works, easily apparent or not. And plenty of inventions are disclosed by being patented, even if they'd probably be able to be kept as trade secrets - for one thing, it is a hedge against someone independently re-discovering it, for another you can license it out to others without having to be paranoid about someone leaking it. The essential element of a patent, though, is still disclosure, and that is the primary way it fulfills the Constitutional mandate to "promote the progress of Science", not by simply granting a monopoly to entice people to invent new things.
Yes, many things are shared openly - open source, scientific papers. It's a good part of an argument that patents may not be necessary to get people to share their ideas. However, that doesn't change the fact that a key element of patents is that they require you to disclose your methods, so that everyone can use them (as in, share!) after the patent period has expired.
Well, you could if you had registered as a bidder in time.
Why would Google care about whether speculative investors buy into the IPO? They'd rather the person who might have bought from the speculative investor buy directly from them. That way Google gets all the money, instead of the speculator skimming some out of the middle.
I don't think you've thought about how ridiculously large 2^128 is. Go back and read what I said. The entire computational capability of the planet, running for the next 10 years, wouldn't even make a noticeable dent in a 2^128 sized space. Those 18,000-processor machines, one for each person, each processor doing femto-second hashes, would take 100 years to do 2^128. How much of a chunk do you think today's multi-nanosecond-per-hash processors would take, even if you had 18,000 of them for each person in the world, and ran them for 10 years? 0.000002 PERCENT, and that's giving a generous 5 nanoseconds per hash. How many 18,000-processor clusters do you think the government actually has access to? Less than 6 billion, I'll bet.
Note also that you're only talking about 16-bit chunks. It doesn't help you with any other sized chunks, of which there are a LOT more than 2^128.
Or, put another way, guessing at the key is identical to guessing at the decrypted message. Just guess what the message might be. You don't even have to intercept the encrypted message to do this. Just guess! "I wonder if it could be ... aha, proof that Iraq has weapons of mass destruction! Wow, that's so unlikely that I randomly guessed that exact message that it must be right. It even has he word "misunderestimate" in it, it must be real!" Maybe the encrypted message is a little short? Guess that they first compressed it with gzip, then added some zero padding (or you can make a guess what the padding is, too), and a signed MD5 signature . Guess what? It's so unlikely that the message just happens to be what you guessed, in gzip format and all, with the exact same padding, and look at that signnature you guessed, that it must be the message. Wow, they managed to encrypt BOTH messages at the same time, what an advanced code!
You do realize how big 2^128 is, don't you?
In storage terms, every person on the Earth today would have to have 56 trillion 16-petabyte storage units to hold that many MD5 hashes. One of those disk drives would hold around 3000 years worth of uncompressed CD-quality stereo music (or only around 450 years of DVD-quality compressed video).
In computational terms, every person would have to have an 18,000-processor machine, each processor doing one hash every femto-second, and it would still take 100 years. The bandwidth required to store the hashes for just one of those machines would be able to transfer around 350 billion CDs per second.
The most ridiculous part is, I could be off by a whole bunch of orders of magnitude and it wouldn't make a difference in how totally off-scale those numbers are.
One problem with that is there isn't any good point that doesn't give a significant response from the blue or the yellow-green cones (not red - there are no red sensors, red is detected by the somewhat higher response of the yellow-green compared to green). Two possibilities for additional colors would be at the ends of sensitivity for green (both ends) and blue (towards the green), and at the points of equal sensitivity (e.g. approximately cyan).
No, even if perfectly calibrated to an individual eye, RGB can't perfectly reproduce real colors, even theoretically. For example, RGB can't create the same response in the eye as indigo or violet light does. Cyan and yellow light can only be approximated by RGB, due to the overlapping of the response curves of the three types of cones.