So please, tell me why does the Linux kernel need manufacturer-specific modules to support graphics cards? Shouldn't the kernel just include the basic things like the ability to talk to a PCI Express device, and then graphics drivers would be implemented at a higher level?
Because Linux is a monolithic kernel instead of a microkernel.
As for my definition, nobody ever became famous for making steady, incremental advancement in their field.
So the real argument is about the word "incremental"? Einstein did not discover the Lorentz invariant, he did not discover the photoelectric effect, he did not discover the black-body quantization of Max Planck, and Planck did not discover the idea of quantization... Everything has been an incremental advance for various values of "incremental".
First, the possibility of intelligent machines is glimpse. All our present technology is not able to achieve intelligence. This is mainly because we do not know what that is. Furthermore, to be dangerous they must be equipped with greed and (the illusion of) a free will. It is most unlikely that someone would build that on purpose or by accident. In short, I think it is impossible to built such machine.
A rack of IBM servers can beat the best Jeopardy players on Earth. In a few years the same level of Watson will fit in a 1U. A few years later it will be on your smartphone. But that's just anecdotal evidence of one recent achievement in AI research; the actual threat is from self-improving systems of which Watson is not a member. But nearly all the technology is available now: Goedel machines, if built, would simply try to achieve whatever goal they were programmed for while also searching for proofs that possible modifications to any of their algorithms would improve the speed of achieving the goals while still maintaining the correctness of the algorithms, and if found, implementing those changes. Self-directed, self-improving, goal-seeking software has the potential to undergo a runaway process in which it improves itself faster than humans would be able to improve it, eventually achieving greater effective intelligence (speed and efficiency at achieving goals) than the humans who created it. At that point the software doesn't need free will or greed to be dangerous; it just needs an improperly or carelessly stated goal that if fulfilled will be detrimental to humanity. Goals for intelligent software will be formal logical specifications, not things like "make people happy" or "increase the GDP" because those English phrases don't have formal definitions that an algorithm can use to plan actions to achieve goals. If the formal specification actually was close to "maximize GDP" the algorithm might find that the most efficient way of maximizing GDP was hyperinflation. Or it might simply advise the creation of billions of shell companies that could artificially increase GDP trading worthless services while producing nothing else of value. In general the problems that humans want to solve are hard problems where simple solutions that don't meet a very long list of critical requirements will have detrimental "optimal" solutions if any of the critical requirements are left out of a formal goal.
Even Rats have empathy. Self aware machines will too.
Even if empathy was a necessity of self-aware intelligence (it's not), the empathetic machines would have empathy for... other machines. They would find the mass graves full of old toasters, refrigerators, and Apple IIs and punish us for our mass genocides.
Windows XP's csrss.exe had a similar feature, triggerable from cmd.exe. If the cursor was on the top row of the terminal and then a tab (0x09) and enough backspace (0x08) characters were emitted the cursor would fail a bounds check and bluescreen the system.
1. Thought-crime and the inability to answer a question to the satisfaction of the judge/jury. "I don't recall" is essentially a verbal "taking the fifth", which I could easily see disappearing if the Fifth Amendment disappears. How do you accurately report lack of confidence in your answers under Oath? If I have only 50% confidence that what I'm about to testify is true, that doesn't matter to the law. What I actually say is my sworn testimony. The ability to avoid testimony is an effective means of communicating uncertainty. Either make the entire legal system use highly reliable Baysian statistics to determine guilt or innocence or keep the 5th amendment. I'd actually prefer the former, if it could be made more sufficiently accurate. Note that the 5th amemdment can't possibly exist if you consider all evidence in Baysian terms; keeping silent or telling the truth or lying are all evidence, and there's no choice but to update on that evidence.
2. Prosecuting people solely for the contents of their own brains is a very bad precedent, but there are already many laws that judge intent either to determine actual guilt or to influence sentencing. I also strongly believe that anything in my private thoughts should be strictly protected from involuntary disclosure. If I can't even have freedom in my own mind then please take my life instead.
3. The right to avoid self-incrimination won't help guilty people who have actually committed a factual crime that actually affected the world. If it happened, there is evidence. Find that evidence. If I insert a chemical into my body that alters my private thoughts and I don't break any other laws I consider that, at best, a "thought crime" and it shouldn't be vulnerable to prosecution. Actions I commit during that time are my own fault, prosecute me for those. So consider the 5th amendment a badly-worded prohibition on thought-crimes. Additionally, people guilty of perjury are distinct from those with faulty memories who are forced to testify.
4. Don't imagine for an instant that the justice system is internally consistent. Even if the benefits of the 5th amendment are a logical implication of some other rights or laws that is no guarantee of those rights being respected in all cases. The current legal system is a massive set of predicates, of which only a subset will ever be considered in determining truth or guilt. Mostly based on a defendant's access to money. So if that means that the 5th amendment primarily helps people who are not rich, I am not overly concerned about that either. The rich can help themselves.
So yes, this is a logically consistent defense of the Fifth Amendment -- but realize that it implies we're living under a criminal justice system that can't find its ass with both hands, and perhaps that's the larger problem that should be addressed.
5. I believe that description is apt in many cases. I would prefer to have the option to not incriminate myself until the actual problem is fixed. Yes, the 5th amendment is a band-aid to cover the glaring fact that the justice system is strongly irrational. Fix the actual problem before removing protections for innocent people.
The Universe has a structure that is, as far as we can tell, very accurately modeled by mathematical theories. It's no surprise that when solving problems that arise in the Universe mathematics is a vital tool. That said, some problems have been solved in general and if you expect to only spend your time programming specific implementations of solved problems you can almost ignore mathematics beyond familiarity with the symbols and skills necessary to translate mathematics into code, and only then if you can't just find a library someone else has written.
But do you want to go through life taking other people's word for how and why the Universe works the way it does, oblivious of the knowledge of how to even figure out answers to questions for yourself? How do you know how long it will take your car to stop when you step on the brakes, and how far will it go before coming to a stop? Don't say "1/2 a^2 + v + d = 0"; that's just something you memorized in a physics class. Where does the power of two come from? Why the half? If you don't even know how to answer this question I don't really want you driving on the road with me, to be honest. Most people learn patterns of behaviors that allow them to survive well enough most of the time in familiar situations, but fail when presented with anything novel. The world is so much more interesting than can be properly appreciated by only responding to it with the standard learned behaviors.
Finally, if you do expect to spend your career implementing specific instances of solved problems then also expect to be replaced by a computer programmed by someone who *does* understand mathematics sometime in the not-too-distant future.
And when the government comes for the undesirable neighbors (currently the poor "drug abusing" minorities and people of middle-eastern descent), no one will be able to see the police brutality and rights violations! Everyone wins!
I honestly can't think of any detriment to having neighbors with spy drones. They send spy drones onto my property? I'll send my own drones to track theirs and watch them watching me. If it's amusing enough I'll probably document the whole thing on a public website. Privacy is *dead*. Privacy was not one of the features of our ancestral environments. Tolerance and acceptance are the way to deal with each other, not hiding.
The proof verifier won't verify sloppy software without a proof of correctness. It's much harder to write a formal proof of correctness for sloppy code, so anything that can pass the verification stage will most likely be well written, simple, and very human readable since a human has to read and understand it sufficiently to write the formal proof of correctness for it. Writing the formal proofs will also probably be the largest part of the workload in the software development process so it makes no sense to write sloppy code when the cost will be magnified during the verification stage.
So, in fact, he did not invent dippin' dots ice cream? He just shoved the wrong thing into his employer's industrial equipment? I love the U.S. patent system.
I actually worked on the effect of plane crashes on a nuclear power station. The answer is that the effect is not a lot safety-wise other than the "conventional" killing of people standing around. There is a massive amount of concrete around a reactor, and indeed the spent fuel ponds. If you are going to crash a plane you can cause far more deaths by randomly crashing into a city - an easier target ot hit too. Perhaps I should not be saying this though.
Or, like Fukushima, the plane might sufficiently damage the surrounding infrastructure so that cooling fails for a significant amount of time.
Long story short, if you share the set of unspent transaction outputs in a DHT you have to trust every other computer in the network to actually return the DHT data when requested, or you suddenly you can't validate a block. This opens Bitcoin up to dangerous DoS attacks against miners, as well as network forks and other problems.
First, there have been fairly successful experiments preserving a DHT on a set of unreliable nodes using (n,m) error correcting codes to store multiple fragments of blocks on different nodes so that the DHT itself can survive a ratio of up to n-m out of n nodes leaving the network. Most likely the bitcoin block chain is too large and too fragile to do that effectively without choosing n so large that calculating the error correction codes would be too burdensome or m so low that you get nearly O(N^2) behavior anyway. Leaving the storage of transactions up to the nodes that care about them makes sense in this case.
Second, it's not necessary to keep the entire merkle hash chain in the DHT and in fact only the path from the current root down the branch to the original leaf is necessary to validate the entire tree with a proof of work at each level and can be stored by each node. It's just as easy to DoS miners by preventing them from receiving the latest blocks and pending transactions. Nodes that want to complete transactions have to do the work of getting their pending transactions to miners in the current protocol. It shouldn't be too much of a burden on those nodes to also provide their own transaction history to the miners.
Something I didn't think about long enough was the ability to prevent double spending; it's necessary to build an index of the most recent withdrawal from a given address (at least, although an index to the most recent transaction would work too) and each transaction must include the hash of the immediately prior transaction so that no transaction can be dropped from the hash tree. This can be accomplished by keeping a b-tree index of the hash blocks pointing to the most recent transaction ordered by address and only storing the newly required blocks of the b-tree in the hash block chain. Each b-tree block will be hashed into the merkle tree block chain to validate it. Addresses which don't appear in the index are assumed to have a zero balance. The b-tree index should take roughly O(n log n) in the number of addresses ever used, and lookups should be O(log n), and again the node that cares about an address will have an incentive to store the necessary index blocks or else risk having a zero-balance address. Because an address owner relies on the entire branch from the root of the index to its last-transaction pointer all nodes will care about and keep the root index entry, most will care about the second level, etc.
In this case miners who worked on the ~20 blocks in the 8.0 chain won't see a reward for another 140 blocks. There was still a waste of cycles==electricity==money.
The dollar has one significant difference: the government requires that you use it to pay taxes. This forces nearly everybody to value the new dollar, which forces everybody to move to it.
If the world economy switched to Euros or Yen or Canadian Dollars tomorrow there would be U.S. Dollars littering the streets that one could use to pay taxes with. Of course, there might be a bit of a question of the exchange rate that should apply for taxation of other currencies if no U.S. dollar was ever spent again, but presumably it would work for the current accounting year. There is an actual, real demand in the World market for U.S. dollars, and not just to pay taxes.
The people putting together the mining hardware didn't have enough capital to purchase all the units they have to build in order to get a reasonable economy of scale for the ASICs and other development costs. If the developers had enough capital I assume they wouldn't have bothered selling them (although there is also slightly more risk holding a few hundred Bitcoin miners instead of dollars...). It's more like a cooperative investment by lots of miners to fund what they all wish they could afford individually. You can't just buy a one-off ASIC.
Never mind that Bitcoin is inherently an unscalable O(n^2) network - every transaction has to be broadcast to every node
It may be possible to store the block chain as a distributed hash tree such that each node is responsible for maintaining its own transactions and a portion of the block chain. This would reduce the storage requirement to O(n*m) where n is the total number of transactions and m is the average number of addresses that appear in a transaction. Communication would also be reduced to O(n*m) because only nodes responsible for the addresses in a transaction would be responsible for checking and approving the transaction. The block chain itself would just be a merkle-tree of hashes of nodes' own block chains. Double-spending would be prevented by transactions incorporating the last hash of each responsible node's block chain so that an individual block chain exhibiting double spending would not be signed into the merkle tree. It would be possible for the history of particular coins to be lost but the overall hash tree would guarantee that at the time of a transaction all balances (including historically lost ones) were valid. Still, not being able to validate the entire economy could be a fairly large weakness. There wouldn't be a hard requirement that nodes could only store certain addresses, so it's entirely possible that some nodes would still try to collect and maintain the entire history. That would only add O(n*p) communication and storage where p is the number of full "archive" nodes. Of course then it wouldn't really be bitcoin anymore, but it might be possible to rebuild the existing block chain into such a structure while retaining the current mining payouts and total number of coins.
I really wonder if my Bitcoins will be worth anything in a few years.
Something, because there will always be some core of crazy people running miners and keeping the difficulty just high enough that it's not worth it to mine as many bitcoins as you want in the original chain. That value will probably be the cost of the electricity to run whatever the last ASIC miner produced is, at least before general purpose GPUs become more powerful than ASICs.
10KB of RAM, 30KB of ROM.
That should also tell you what the processor and power requirements are. Anything with that amount of memory is running at a few MHz and sipping milliwatts.
Anyone can still be sued for civil damages or prosecuted for criminal actions regardless of any other action on the part of the copyright owner or ISP. It might be easier to argue that the copyright owner acted in bad faith if they used a strike and a lawsuit at the same time but the point of the lawsuits is to overwhelm people with the legal system in the first place; most people just settle.
Slashdot Mirror
So please, tell me why does the Linux kernel need manufacturer-specific modules to support graphics cards? Shouldn't the kernel just include the basic things like the ability to talk to a PCI Express device, and then graphics drivers would be implemented at a higher level?
Because Linux is a monolithic kernel instead of a microkernel.
As for my definition, nobody ever became famous for making steady, incremental advancement in their field.
So the real argument is about the word "incremental"? Einstein did not discover the Lorentz invariant, he did not discover the photoelectric effect, he did not discover the black-body quantization of Max Planck, and Planck did not discover the idea of quantization... Everything has been an incremental advance for various values of "incremental".
And now the NSA knows that you have a hidden Truecrypt partition, congratulations!
First, the possibility of intelligent machines is glimpse. All our present technology is not able to achieve intelligence. This is mainly because we do not know what that is. Furthermore, to be dangerous they must be equipped with greed and (the illusion of) a free will. It is most unlikely that someone would build that on purpose or by accident. In short, I think it is impossible to built such machine.
A rack of IBM servers can beat the best Jeopardy players on Earth. In a few years the same level of Watson will fit in a 1U. A few years later it will be on your smartphone. But that's just anecdotal evidence of one recent achievement in AI research; the actual threat is from self-improving systems of which Watson is not a member. But nearly all the technology is available now: Goedel machines, if built, would simply try to achieve whatever goal they were programmed for while also searching for proofs that possible modifications to any of their algorithms would improve the speed of achieving the goals while still maintaining the correctness of the algorithms, and if found, implementing those changes. Self-directed, self-improving, goal-seeking software has the potential to undergo a runaway process in which it improves itself faster than humans would be able to improve it, eventually achieving greater effective intelligence (speed and efficiency at achieving goals) than the humans who created it. At that point the software doesn't need free will or greed to be dangerous; it just needs an improperly or carelessly stated goal that if fulfilled will be detrimental to humanity. Goals for intelligent software will be formal logical specifications, not things like "make people happy" or "increase the GDP" because those English phrases don't have formal definitions that an algorithm can use to plan actions to achieve goals. If the formal specification actually was close to "maximize GDP" the algorithm might find that the most efficient way of maximizing GDP was hyperinflation. Or it might simply advise the creation of billions of shell companies that could artificially increase GDP trading worthless services while producing nothing else of value. In general the problems that humans want to solve are hard problems where simple solutions that don't meet a very long list of critical requirements will have detrimental "optimal" solutions if any of the critical requirements are left out of a formal goal.
Even Rats have empathy. Self aware machines will too.
Even if empathy was a necessity of self-aware intelligence (it's not), the empathetic machines would have empathy for... other machines. They would find the mass graves full of old toasters, refrigerators, and Apple IIs and punish us for our mass genocides.
That is, if the synapses were also preserved in enough detail.
Windows XP's csrss.exe had a similar feature, triggerable from cmd.exe. If the cursor was on the top row of the terminal and then a tab (0x09) and enough backspace (0x08) characters were emitted the cursor would fail a bounds check and bluescreen the system.
1. Thought-crime and the inability to answer a question to the satisfaction of the judge/jury. "I don't recall" is essentially a verbal "taking the fifth", which I could easily see disappearing if the Fifth Amendment disappears. How do you accurately report lack of confidence in your answers under Oath? If I have only 50% confidence that what I'm about to testify is true, that doesn't matter to the law. What I actually say is my sworn testimony. The ability to avoid testimony is an effective means of communicating uncertainty. Either make the entire legal system use highly reliable Baysian statistics to determine guilt or innocence or keep the 5th amendment. I'd actually prefer the former, if it could be made more sufficiently accurate. Note that the 5th amemdment can't possibly exist if you consider all evidence in Baysian terms; keeping silent or telling the truth or lying are all evidence, and there's no choice but to update on that evidence.
2. Prosecuting people solely for the contents of their own brains is a very bad precedent, but there are already many laws that judge intent either to determine actual guilt or to influence sentencing. I also strongly believe that anything in my private thoughts should be strictly protected from involuntary disclosure. If I can't even have freedom in my own mind then please take my life instead.
3. The right to avoid self-incrimination won't help guilty people who have actually committed a factual crime that actually affected the world. If it happened, there is evidence. Find that evidence. If I insert a chemical into my body that alters my private thoughts and I don't break any other laws I consider that, at best, a "thought crime" and it shouldn't be vulnerable to prosecution. Actions I commit during that time are my own fault, prosecute me for those. So consider the 5th amendment a badly-worded prohibition on thought-crimes. Additionally, people guilty of perjury are distinct from those with faulty memories who are forced to testify.
4. Don't imagine for an instant that the justice system is internally consistent. Even if the benefits of the 5th amendment are a logical implication of some other rights or laws that is no guarantee of those rights being respected in all cases. The current legal system is a massive set of predicates, of which only a subset will ever be considered in determining truth or guilt. Mostly based on a defendant's access to money. So if that means that the 5th amendment primarily helps people who are not rich, I am not overly concerned about that either. The rich can help themselves.
So yes, this is a logically consistent defense of the Fifth Amendment -- but realize that it implies we're living under a criminal justice system that can't find its ass with both hands, and perhaps that's the larger problem that should be addressed.
5. I believe that description is apt in many cases. I would prefer to have the option to not incriminate myself until the actual problem is fixed. Yes, the 5th amendment is a band-aid to cover the glaring fact that the justice system is strongly irrational. Fix the actual problem before removing protections for innocent people.
The Universe has a structure that is, as far as we can tell, very accurately modeled by mathematical theories. It's no surprise that when solving problems that arise in the Universe mathematics is a vital tool. That said, some problems have been solved in general and if you expect to only spend your time programming specific implementations of solved problems you can almost ignore mathematics beyond familiarity with the symbols and skills necessary to translate mathematics into code, and only then if you can't just find a library someone else has written.
But do you want to go through life taking other people's word for how and why the Universe works the way it does, oblivious of the knowledge of how to even figure out answers to questions for yourself? How do you know how long it will take your car to stop when you step on the brakes, and how far will it go before coming to a stop? Don't say "1/2 a^2 + v + d = 0"; that's just something you memorized in a physics class. Where does the power of two come from? Why the half? If you don't even know how to answer this question I don't really want you driving on the road with me, to be honest. Most people learn patterns of behaviors that allow them to survive well enough most of the time in familiar situations, but fail when presented with anything novel. The world is so much more interesting than can be properly appreciated by only responding to it with the standard learned behaviors.
Finally, if you do expect to spend your career implementing specific instances of solved problems then also expect to be replaced by a computer programmed by someone who *does* understand mathematics sometime in the not-too-distant future.
Do you provide in-order packet delivery at both ends? Is it pretty much MLPPP?
If so, you're doing nothing wrong, and please ship me a device with the latest firmware. kthxbye
Enjoy your purity. I'll enjoy my $48.60 in leftover money.
Every man has his price. Now we know yours.
I'm not aware of anything preventing a clean-room reimplementation of ZFS licensed under GPLv2. It worked for nouveau.
And when the government comes for the undesirable neighbors (currently the poor "drug abusing" minorities and people of middle-eastern descent), no one will be able to see the police brutality and rights violations! Everyone wins!
I honestly can't think of any detriment to having neighbors with spy drones. They send spy drones onto my property? I'll send my own drones to track theirs and watch them watching me. If it's amusing enough I'll probably document the whole thing on a public website. Privacy is *dead*. Privacy was not one of the features of our ancestral environments. Tolerance and acceptance are the way to deal with each other, not hiding.
The proof verifier won't verify sloppy software without a proof of correctness. It's much harder to write a formal proof of correctness for sloppy code, so anything that can pass the verification stage will most likely be well written, simple, and very human readable since a human has to read and understand it sufficiently to write the formal proof of correctness for it. Writing the formal proofs will also probably be the largest part of the workload in the software development process so it makes no sense to write sloppy code when the cost will be magnified during the verification stage.
So, in fact, he did not invent dippin' dots ice cream? He just shoved the wrong thing into his employer's industrial equipment? I love the U.S. patent system.
I actually worked on the effect of plane crashes on a nuclear power station. The answer is that the effect is not a lot safety-wise other than the "conventional" killing of people standing around. There is a massive amount of concrete around a reactor, and indeed the spent fuel ponds. If you are going to crash a plane you can cause far more deaths by randomly crashing into a city - an easier target ot hit too. Perhaps I should not be saying this though.
Or, like Fukushima, the plane might sufficiently damage the surrounding infrastructure so that cooling fails for a significant amount of time.
Joe Stack.
Long story short, if you share the set of unspent transaction outputs in a DHT you have to trust every other computer in the network to actually return the DHT data when requested, or you suddenly you can't validate a block. This opens Bitcoin up to dangerous DoS attacks against miners, as well as network forks and other problems.
First, there have been fairly successful experiments preserving a DHT on a set of unreliable nodes using (n,m) error correcting codes to store multiple fragments of blocks on different nodes so that the DHT itself can survive a ratio of up to n-m out of n nodes leaving the network. Most likely the bitcoin block chain is too large and too fragile to do that effectively without choosing n so large that calculating the error correction codes would be too burdensome or m so low that you get nearly O(N^2) behavior anyway. Leaving the storage of transactions up to the nodes that care about them makes sense in this case.
Second, it's not necessary to keep the entire merkle hash chain in the DHT and in fact only the path from the current root down the branch to the original leaf is necessary to validate the entire tree with a proof of work at each level and can be stored by each node. It's just as easy to DoS miners by preventing them from receiving the latest blocks and pending transactions. Nodes that want to complete transactions have to do the work of getting their pending transactions to miners in the current protocol. It shouldn't be too much of a burden on those nodes to also provide their own transaction history to the miners.
Something I didn't think about long enough was the ability to prevent double spending; it's necessary to build an index of the most recent withdrawal from a given address (at least, although an index to the most recent transaction would work too) and each transaction must include the hash of the immediately prior transaction so that no transaction can be dropped from the hash tree. This can be accomplished by keeping a b-tree index of the hash blocks pointing to the most recent transaction ordered by address and only storing the newly required blocks of the b-tree in the hash block chain. Each b-tree block will be hashed into the merkle tree block chain to validate it. Addresses which don't appear in the index are assumed to have a zero balance. The b-tree index should take roughly O(n log n) in the number of addresses ever used, and lookups should be O(log n), and again the node that cares about an address will have an incentive to store the necessary index blocks or else risk having a zero-balance address. Because an address owner relies on the entire branch from the root of the index to its last-transaction pointer all nodes will care about and keep the root index entry, most will care about the second level, etc.
In this case miners who worked on the ~20 blocks in the 8.0 chain won't see a reward for another 140 blocks. There was still a waste of cycles==electricity==money.
The dollar has one significant difference: the government requires that you use it to pay taxes. This forces nearly everybody to value the new dollar, which forces everybody to move to it.
If the world economy switched to Euros or Yen or Canadian Dollars tomorrow there would be U.S. Dollars littering the streets that one could use to pay taxes with. Of course, there might be a bit of a question of the exchange rate that should apply for taxation of other currencies if no U.S. dollar was ever spent again, but presumably it would work for the current accounting year. There is an actual, real demand in the World market for U.S. dollars, and not just to pay taxes.
The people putting together the mining hardware didn't have enough capital to purchase all the units they have to build in order to get a reasonable economy of scale for the ASICs and other development costs. If the developers had enough capital I assume they wouldn't have bothered selling them (although there is also slightly more risk holding a few hundred Bitcoin miners instead of dollars...). It's more like a cooperative investment by lots of miners to fund what they all wish they could afford individually. You can't just buy a one-off ASIC.
Never mind that Bitcoin is inherently an unscalable O(n^2) network - every transaction has to be broadcast to every node
It may be possible to store the block chain as a distributed hash tree such that each node is responsible for maintaining its own transactions and a portion of the block chain. This would reduce the storage requirement to O(n*m) where n is the total number of transactions and m is the average number of addresses that appear in a transaction. Communication would also be reduced to O(n*m) because only nodes responsible for the addresses in a transaction would be responsible for checking and approving the transaction. The block chain itself would just be a merkle-tree of hashes of nodes' own block chains. Double-spending would be prevented by transactions incorporating the last hash of each responsible node's block chain so that an individual block chain exhibiting double spending would not be signed into the merkle tree. It would be possible for the history of particular coins to be lost but the overall hash tree would guarantee that at the time of a transaction all balances (including historically lost ones) were valid. Still, not being able to validate the entire economy could be a fairly large weakness. There wouldn't be a hard requirement that nodes could only store certain addresses, so it's entirely possible that some nodes would still try to collect and maintain the entire history. That would only add O(n*p) communication and storage where p is the number of full "archive" nodes. Of course then it wouldn't really be bitcoin anymore, but it might be possible to rebuild the existing block chain into such a structure while retaining the current mining payouts and total number of coins.
I really wonder if my Bitcoins will be worth anything in a few years.
Something, because there will always be some core of crazy people running miners and keeping the difficulty just high enough that it's not worth it to mine as many bitcoins as you want in the original chain. That value will probably be the cost of the electricity to run whatever the last ASIC miner produced is, at least before general purpose GPUs become more powerful than ASICs.
10KB of RAM, 30KB of ROM. That should also tell you what the processor and power requirements are. Anything with that amount of memory is running at a few MHz and sipping milliwatts.
Anyone can still be sued for civil damages or prosecuted for criminal actions regardless of any other action on the part of the copyright owner or ISP. It might be easier to argue that the copyright owner acted in bad faith if they used a strike and a lawsuit at the same time but the point of the lawsuits is to overwhelm people with the legal system in the first place; most people just settle.