If even 1/10 of the hype about "breakthroughs" in solar cell efficiency were actually to be combined and made real in the marketplace, we'd all be charging the utility companies now instead of the other way around.
"Petaflops" is not representative of the power of modern supercomputers, many of which use massively parallel integer processing to perform their duties. Sure, you can say that simulating floating point operations with the integer units amounts to the same thing, but it actually doesn't. We have discovered that there are a great many real-world problems for which parallel integer math works just fine, or even better (more efficient) than floating point. And for those, flops is a completely meaningless metric.
Unlike the "many other industries" you mention, in the case of Hashrocket, it is a company that (in their own view) is looking for the very best and brightest.
Well, guess what? If you want the best and brightest, you are going to have to pay for them. You are going to miss a shitload of them if you make them invest a full week's time and money, plus lost wages, just to be considered.
But the fact remains that an entire week is a rather large investment to ask on the part of the applicant, with no compensation, and regardless of plane tickets and other accommodation, probably at significant expense as well.
In addition to the fact that it appears to be operating right on the borderline of legality, or perhaps even over it. Exactly where it lies in that regard, though, is a matter of state law, so it's hard to say.
"Not quite. You also need wires to connect the NANDs together, and those wires must be capable of fanout, so really, it's NAND gates + wires + wire junctions."
Well, that's true. The gates need a way to communicate with each other. But in most cases the "wires" are a trivial matter. In Conway's "Life", information transfer took the form of "gliders" that could travel from "gate" to "gate".
"But raising your ask over their ask will just as surely not result in a trade (because if there is any trade it will be at the lower ask)."
In an idea market that would be true, but if the market were actually ideal no trading would ever take place.
In order for a stock to rise in price, somebody must pay more than the current market price. When I wrote "price" I did mean ask, but the whole purpose of HFT is to take advantage of minute fluctuations... where somebody is paying more than somebody else. If those fluctuations did not occur, there would be no market. If nobody ever paid more than the current market price, stocks could only remain flat or go down, never up.
"But the cause is not nearly so simplistic as you imagine."
YOU are the one misunderstanding here. I did not accuse HFT of causing bubbles. What I *DID* write, however, is that if the current trends and regulations continue, such a bubble in inevitable. NOT because of the way it normally affects the market, but due to software error.
As I said, it has already happened once on Amazon, in which books that normally sold in the 2-digit range had their prices suddenly raised over a few days to 7 digits. Obviously the books -- like any stock that happened to -- would not sell at that price. But loops can happen in the other direction, too. Which is only a very small example, and that was a direct loop: two vying systems going at each other. But indirect loops are possible too and no amount of software cleverness is likely to completely eliminate them. Remember that recent $440 million loss?
Obviously smart software should be aware of such things, but with all the HFT going on, it's not a case of just two parties. There is very real danger that an indirect recursion could happen, in ways the software won't detect.
No, it doesn't. That's all fine when stock trading is done in the slow, traditional way, and values go up over time. It isn't the case when you are doing "High Frequency". In essence, it looks for any small advantage and takes it.
In other words: you often have to assume that somebody out there is willing to pay just a fraction more than you did, pretty soon. So you raise your price 1/1000th of a penny higher than the next guy's.
But somebody else is doing the same thing. And their price goes up 1/1000th of a cent higher than yours. And so on. Back and forth like that half a million times, and your stock price is now up $10 over what the rest of the market will pay.
Don't laugh. This happened on Amazon recently, when two book sellers used software to set their prices just the tiniest smidgin higher than the next highest price. After a few days the books had a list price of well over $1 million.
"It basically means that as far as possible, to have proven source acquisition paths for every file on your computer."
This is really beside the point, though. Even if it is a civil matter, the burden to show at least "a preponderance of evidence" is supposed to be on the plaintiff (or complainant, if you prefer). Having to prove your innocence is not exactly a traditional American concept.
"regardless of my personal convictions, it looks like torrenting songs is a big crime with big consequences."
No, it isn't. In the vast majority of cases, it isn't a crime at all, not even a misdemeanor. It is a civil violation. But it can still have harsh consequences.
Of course, the statutory damages are far too high to provide anything like justice; the amount should be based on actual damages.
"The problem is that there is a penalty of "up to $150,000" for copying a song, which would probably appropriate if let's say a day after Michael Jackson's death a record company had released a CD "Michael Jackson's greatest hits" without owning the copyrights, and sold millions of that."
Nope. Wrong.
Downloading and other infringement for personal use is a misdemeanor with the "up to $150,000" fine, which I think is very excessive. But it is nevertheless a civil infraction, not a crime.
But illegally selling millions of copies of a copyrighted work is a felony, with far harsher punishment. They are two very different laws.
"Historically, jury nullification freed the Klansman and hanged the black man."
There is A LOT more to its history than that. Has it been used for what most of us consider to be bad purposes? Sure. Far more often, it has been used to nullify bad or unconstitutional laws.
Go visit and read some of their material about the history of nullification.
"No they are not allowed to tell the jury about jury nullification. technically no one is allowed to tell the jury about jury nullification, and doing do would be precedent for a mistrial."
Yes, they are allowed. The judge is the one who ultimately decides. I happen to know the Executive Director of The Fully Informed Jury Association, and I am familiar with the political and legal history of jury nullification. The actual situation is different than you present.
Many judges and prosecutors don't like the jury to know about jury nullification, because they don't like the jury to know that in many ways it has more power than they do. But they can and in some cases have told the jury about it themselves.
"You keep citing that web page but I'm pretty sure you haven't read it."
I'm pretty sure YOU haven't read it, because it's far more than just one page.
The part I was specifically referring to was the part that outlaws interception of cell phone signals... which are radio signals, often (usually) unencrypted, over the public airwaves.
I am not an HF trader but I am a software engineer, and regulation is not going to help very much in this field. Sooner or later two or more firms are going to get caught in an indirect loop, causing a destructive price war, or something of that nature. It's pretty much inevitable. In fact it has already happened, but only in small degree.
You can't regulate around that unless you try to regulate every line of code, and I think we all know that won't work either.
"The Federal Reserve would more properly be called the Third Bank of the United States."
Not really. It would more properly be called the Third Reserve. There is nothing Federal about it, except that the Chairman of the Board is appointed by the President. It does not "belong" to the United States. The Fed is a collection of private banks, including a great deal of foreign interest.
"If you round to the nearest millionth of a penny then millions of dollars of infrastructure will only bring in fractional millionths of a penny..."
This is not quite accurate. I'm not sure, but I think you meant trading at the level of millionths of a penny price difference. They are not the same thing. You can take any amount at all and round it to a millionth of a penny. That would make absolutely no difference.
Slashdot Mirror
If even 1/10 of the hype about "breakthroughs" in solar cell efficiency were actually to be combined and made real in the marketplace, we'd all be charging the utility companies now instead of the other way around.
"Petaflops" is not representative of the power of modern supercomputers, many of which use massively parallel integer processing to perform their duties. Sure, you can say that simulating floating point operations with the integer units amounts to the same thing, but it actually doesn't. We have discovered that there are a great many real-world problems for which parallel integer math works just fine, or even better (more efficient) than floating point. And for those, flops is a completely meaningless metric.
We need a standard that actually makes sense.
I meant to add:
Unlike the "many other industries" you mention, in the case of Hashrocket, it is a company that (in their own view) is looking for the very best and brightest.
Well, guess what? If you want the best and brightest, you are going to have to pay for them. You are going to miss a shitload of them if you make them invest a full week's time and money, plus lost wages, just to be considered.
That's the way it works.
Do I smell shill?
Just kidding!
But the fact remains that an entire week is a rather large investment to ask on the part of the applicant, with no compensation, and regardless of plane tickets and other accommodation, probably at significant expense as well.
In addition to the fact that it appears to be operating right on the borderline of legality, or perhaps even over it. Exactly where it lies in that regard, though, is a matter of state law, so it's hard to say.
In any case, it rather strains my idea of ethics.
Haven't seen you around in a while. You probably have been, but I haven't seen you much.
"Not quite. You also need wires to connect the NANDs together, and those wires must be capable of fanout, so really, it's NAND gates + wires + wire junctions."
Well, that's true. The gates need a way to communicate with each other. But in most cases the "wires" are a trivial matter. In Conway's "Life", information transfer took the form of "gliders" that could travel from "gate" to "gate".
"But raising your ask over their ask will just as surely not result in a trade (because if there is any trade it will be at the lower ask)."
In an idea market that would be true, but if the market were actually ideal no trading would ever take place.
In order for a stock to rise in price, somebody must pay more than the current market price. When I wrote "price" I did mean ask, but the whole purpose of HFT is to take advantage of minute fluctuations... where somebody is paying more than somebody else. If those fluctuations did not occur, there would be no market. If nobody ever paid more than the current market price, stocks could only remain flat or go down, never up.
"But the cause is not nearly so simplistic as you imagine."
YOU are the one misunderstanding here. I did not accuse HFT of causing bubbles. What I *DID* write, however, is that if the current trends and regulations continue, such a bubble in inevitable. NOT because of the way it normally affects the market, but due to software error.
As I said, it has already happened once on Amazon, in which books that normally sold in the 2-digit range had their prices suddenly raised over a few days to 7 digits. Obviously the books -- like any stock that happened to -- would not sell at that price. But loops can happen in the other direction, too. Which is only a very small example, and that was a direct loop: two vying systems going at each other. But indirect loops are possible too and no amount of software cleverness is likely to completely eliminate them. Remember that recent $440 million loss?
If you can build NAND gates (or NOR gates for that matter) out of your materials, you have everything you need to build a Turing-complete machine.
... because it's kind of a pointless exercise. After all, Conway's "Life" game was turing complete, and that had like, what? Maybe 3 rules?
Darn it, I hit "submit" too soon.
Obviously smart software should be aware of such things, but with all the HFT going on, it's not a case of just two parties. There is very real danger that an indirect recursion could happen, in ways the software won't detect.
"Buy low, sell high" prevents that recursion."
No, it doesn't. That's all fine when stock trading is done in the slow, traditional way, and values go up over time. It isn't the case when you are doing "High Frequency". In essence, it looks for any small advantage and takes it.
In other words: you often have to assume that somebody out there is willing to pay just a fraction more than you did, pretty soon. So you raise your price 1/1000th of a penny higher than the next guy's.
But somebody else is doing the same thing. And their price goes up 1/1000th of a cent higher than yours. And so on. Back and forth like that half a million times, and your stock price is now up $10 over what the rest of the market will pay.
Don't laugh. This happened on Amazon recently, when two book sellers used software to set their prices just the tiniest smidgin higher than the next highest price. After a few days the books had a list price of well over $1 million.
"It basically means that as far as possible, to have proven source acquisition paths for every file on your computer."
This is really beside the point, though. Even if it is a civil matter, the burden to show at least "a preponderance of evidence" is supposed to be on the plaintiff (or complainant, if you prefer). Having to prove your innocence is not exactly a traditional American concept.
"regardless of my personal convictions, it looks like torrenting songs is a big crime with big consequences."
No, it isn't. In the vast majority of cases, it isn't a crime at all, not even a misdemeanor. It is a civil violation. But it can still have harsh consequences.
Of course, the statutory damages are far too high to provide anything like justice; the amount should be based on actual damages.
Pardon me. "Misdemeanor" is not correct. As I stated afterward, it is a civil infraction, not a crime. Not even a misdemeanor.
"The problem is that there is a penalty of "up to $150,000" for copying a song, which would probably appropriate if let's say a day after Michael Jackson's death a record company had released a CD "Michael Jackson's greatest hits" without owning the copyrights, and sold millions of that."
Nope. Wrong.
Downloading and other infringement for personal use is a misdemeanor with the "up to $150,000" fine, which I think is very excessive. But it is nevertheless a civil infraction, not a crime.
But illegally selling millions of copies of a copyrighted work is a felony, with far harsher punishment. They are two very different laws.
"They are common law judges. Their hands are never really tied."
This is not at all true. The judges had to try on the basis of very clear federal legislation, which set the statutory damages.
They might have some leeway, but they can't contradict Congress' clear intent unless the law is unconstitutional.
I don't agree with the law, but Congress did make the law.
"Historically, jury nullification freed the Klansman and hanged the black man."
There is A LOT more to its history than that. Has it been used for what most of us consider to be bad purposes? Sure. Far more often, it has been used to nullify bad or unconstitutional laws.
Go visit and read some of their material about the history of nullification.
"No they are not allowed to tell the jury about jury nullification. technically no one is allowed to tell the jury about jury nullification, and doing do would be precedent for a mistrial."
Yes, they are allowed. The judge is the one who ultimately decides. I happen to know the Executive Director of The Fully Informed Jury Association, and I am familiar with the political and legal history of jury nullification. The actual situation is different than you present.
Many judges and prosecutors don't like the jury to know about jury nullification, because they don't like the jury to know that in many ways it has more power than they do. But they can and in some cases have told the jury about it themselves.
"Lol. You've now given away what state you live in."
Haha. Not bloody likely. I got that information from Wikipedia.
"You keep citing that web page but I'm pretty sure you haven't read it."
I'm pretty sure YOU haven't read it, because it's far more than just one page.
The part I was specifically referring to was the part that outlaws interception of cell phone signals... which are radio signals, often (usually) unencrypted, over the public airwaves.
Gee... just like wifi.
I have to agree with eldavajohn.
I am not an HF trader but I am a software engineer, and regulation is not going to help very much in this field. Sooner or later two or more firms are going to get caught in an indirect loop, causing a destructive price war, or something of that nature. It's pretty much inevitable. In fact it has already happened, but only in small degree.
You can't regulate around that unless you try to regulate every line of code, and I think we all know that won't work either.
"The Federal Reserve would more properly be called the Third Bank of the United States."
Not really. It would more properly be called the Third Reserve. There is nothing Federal about it, except that the Chairman of the Board is appointed by the President. It does not "belong" to the United States. The Fed is a collection of private banks, including a great deal of foreign interest.
"If you round to the nearest millionth of a penny then millions of dollars of infrastructure will only bring in fractional millionths of a penny..."
This is not quite accurate. I'm not sure, but I think you meant trading at the level of millionths of a penny price difference. They are not the same thing. You can take any amount at all and round it to a millionth of a penny. That would make absolutely no difference.
It's possible, but you don't know this guy. I don't see that being a problem unless they have pretty weird personalities or culture.
We all have our preferences. If this were about PHP or Java or something, you would likely already have been modded as flamebait.