It's hard to google pre-heartbleed OpenSSL flaws, but there were some serious, subtle flaws in OpenSSL that looked remarkably like typos. After the NSA leaks, there's no doubt: someone committed those flaws deliberately. And the NSA leaks showed a large and well-funded program to do just that: to subvert every public cryptographic tool and standard in subtle ways, vulnerabilities that left tools secure unless you knew about the backdoor (which is particularly pernicious, as when the backdoor is inevitably discovered, the tools are in widespread use).
The open/closed source debate is like a school yard brawl when the Marines land - entirely trumped by vastly more resources spent subverting the tools than went in to writing them. What a damn waste.
With open source, you can start making your own version and modifications
That is the one real advantage. It's not cheap or easy. It's not going to be a hobby project. But it's possible.
The replacement of OpenSSL, the TrueCrypt audit and fork. That's where you see open source step ahead.
We already have countless pieces of evidence of companies being in bed with the government
There's a big difference between a company giving data to the government -- security doesn't enter into that -- and adding deliberate flaws to security products. There hasn't been much evidence of the latter, though wasn't RSA tainted? The bigger worry with proprietary security products is that they're scams, and that happens a lot, but that's a different issue.
Are you looking at the code? I don't think that's relevant.
Companies like Google, Apple, and yes Microsoft have plenty of smart people looking at their closed code for security flaws - well-trained people who's day job is to do just that.
The once-believed advantage of open-source was that companies might be in bed with the NSA, putting flaws in deliberately, but open-source projects wouldn't be. Turns out, not so much. Both groups are just as vulnerable to malicious insiders, and both are filled with techies who would be quite angry to discover a flaw deliberately hidden in their codebase.
You seem to be confusing "tech worker" and "software developer". Be the latter. Even as a software developer, first-job pay will be crap, but who cares. Once you demonstrate you can actually code professionally, it's among the best-paying jobs in America.
It's not about the obvious backdoor. It's often about the random number generator used for generating keys. Maybe that keyspace is smaller than you think.
How many of the e.g.cyanogenmod people collect a paycheck from the NSA? We've seen very subtle flaws in open source code that looked plausibly like a typo, but weakens security just enough for a powerful attacker while remaining secure from a script kiddy.
Not like it's just open source. Trust was lost for the hardware RNG in Intel CPUs (I'm not sure there was ever any evidence of tampering, only evidence of how subtly it could be done: only one guy messing with a mask at the last minute, and the RNG output would still look random).
Well, good luck to you, but whatever you're doing doesn't seem to be making you happy. For sure, if you believe there's no winning strategy, you're not going to find one. Don't talk yourself into hopelessness.
Are you seriously surrounded by people who mow their own yards? Can you even buy a car with no airbags? Are you tired of unlimited free porn? Do you mistake a down business cycle for a non-cyclic trend?
The world didn't end after the 30s, or the 70s, and it won't end after the past 10 years either. Whatever life choices you've made that have left you so pessimistic and unhappy (and I'm guessing unsuccessful?), maybe try different choices?
What, a time before Crocknight lied every night about the Vietnam War because he wanted us out of it? Before the a CIA agent who wanted Nixon out used a of couple newspaper reporters as tools to accomplish his political agenda? Before 5 steady years of wartime propaganda to keep morale up on the home front? Before the press conspired to hide the fact the president was a Polio victim who couldn't walk? Was there ever actually a golden age when the press wasn't just politics? I doubt it. Oh, the political agenda changes from generation to generation, but that's about it.
The only part of the paper you can believe is the sports section. I doubt it's ever been different.
But you can declare, in your docs, what accuracy you promise, and then either succeed or fail at that. And that's the issue here, no? Intel fails at their own published accuracy.
How is there no "correct answer" to e.g cos(0.4) to some specified accuracy? It's right or it's not. Sure, there may be some undefined bits if the precision of the result exceeds the required accuracy, but to whatever accuracy was promised, the result is correct or not.
Sony? Really? The Justice Department bitchslapped Sony so hard the shareholders cried (they missed their quarter, by a lot, due to the fines and the CEO resigned) over the rootkit fiasco. There are real examples of corporate corruption of the justice system, but that's not one of them.
That reasoning is too specific to this one experiment - you always get that cos^2 out of the QM for cases like this. And the chance that a single photon passes through really does go by cos^2 - that's the lesson here, IMO, that energy isn't just an accident that we happen to be squaring something before we measure, but instead that it's the common pattern we observe across both QM and relativity. There's always some sort of vector projection and then squaring going on, and while what the vector is/measures changes from system to system, the square of it keeps looking like some kind of energy.
Heck, given how well Hamiltonians and Lagrangians work in explaining mechanics, I'm starting to think of force as an accidental by-product of potential energy changing over distance (which again makes sense across both QM and GR).
But then that's the real lesson here: that what our intuitions tell us are the fundamental parts of physics from which other stuff should be deduced all comes from human-scale interaction. Pretty silly to think that would be right, given that's not at all the scale at which the fundamental interactions are happening.
You do some optimization on code as you write, but with AI code you can miss big, leave n^2 stuff in that your abstraction layers hid from you. It's not like you were ignoring performance when your wrote it, but when you're trying to get the functionality write you're looking at a tall pile of abstraction.
Once it works, there's often huge wins available from care refactoring for performance. But it's shortchanging the guy who wrote it to say "no one's ever optimized it".
You have a baseball bat at a random angle in the same plane as a picket fence. The spacing on the pickets is such that the bat will fit through if it's within 45% of vertical (or whatever direction the pickets go), and won't fit through at a greater angle. If the distribution of bat angles is even, what's the odds that a bat that passes through the first fence pass though a second at an angle of theta to the first?
It falls off directly with the angular difference. At a theta of 45 degrees, half of possible bats that fit through the first also fit through the second. At a theta of 60 degrees, one third of bats that fit through the first fit through the second. It's the same as taking two quarter-circles, and finding the overlapping length of arc when they're offset by theta (area/length of arc is where "proportional to the angular difference" would come from), as each unit-length of arc represents an equal change of encountering a bat at that range of angles.
That's the wrong answer, of course. Also, any model by which the first fence modifies the photon-substitute in some way gets very odd for the case of entangled pairs of photons, as then you need spooky action at a distance (so that modifying one instantly modifies the other, making relativity cry), and could easily get the wrong answer for 3 filters in a row.
Vector projection would give cos(theta), of course, which is also wrong, unless I misunderstand what you're saying. It's only the square of the vector projection, an energy measurement, that gets you there. I've never met anyone who would have guessed cos^2 ahead of time, but it doesn't really seem that strange that it should be energy that matters once you figure it out - why not energy, after all? (And the energy metaphor works well later on, if you accept the chance that a particle might exist in a certain place as a form of potential energy, in an energy-mass equivalence sort of way).
Anyhow, all of this seems very teachable to high-school students, and simple experiments should establish that light behaves neither light a wave nor like a particle, but "behaves in its own inimical way".
If science works, then we can accept scientific research that altering brain chemistry has a powerful affect on our ability to think rationally, and not for the better. There's no ESP to activate, no magical organ to access, there's just misbehaving wetware if the chemistry gets screwed up (or, if you have evidence otherwise, James Randi has $1 M for you).
Judgment is important. The sense of what's important and what's not is, after all, going to influence your success in life greatly, however you might measure that. If you're acting effectively at random, you're unlikely to get vey far, and focusing on the wrong things to spend your mental effort on is nearly as bad.
It's possible to imagine a drug that would give you deep insight into reality, not mere entertainment. But defeating your judgment such that you mistaken assign deep meaning to random sensation is not that, however entertaining it might be.
True enough, but it's a compelling illusion. Stuff that would dismissed after a college late-night rambling discussion of everything gets instead taken as a profound insight into the workings of the universe. But the drug didn't give you any more insight than you had before, it merely made you think your conclusion was more weighty than it was.
Two-earner households aren't that old. While women went to work during WWII, they stopped when the men came back home.
On the other hand, the automation that the Luddites fought against caused the better part of a century's worth of economic hardship, but it did eventually recover, so maybe it is just a dip long enough to last generations-worth of working lives.
You have an odd view of history if you think the industrial revolution was a bad thing for our standard of living, but one way or another this will be the same sort. I think you seriously underestimate the benefit of technology on our daily lives (technology: the ability to produce a good or deliver a service more efficiently - consuming less labor, or less power, or less raw materials, or all of these). It's what enables exponential growth of the economy at constant population.
After all, the average standard of living is simply all the goods and services all of us produce, divided by the population. Technology is really the only way to drive that average up, if you think about it. And while "net worth" may have very unequal distribution, goods and services are remarkably equal these days, at least in the West, if you ignore social status markers in the mix. Stuff made by robots isn't just going to sit in a warehouse unused - one way or another, there will be that much more goods and services distributed among all of us, and exponential growth stomps minor variations in distribution of stuff, given time.
It's not vector projection in any straightforward way. With the most obvious mental model of polarization (light pointed more in the direction of the filter than perpendicular to it gets through), you expect the energy passing through both filters to fall off directly proportional to the angle between them. If you just took the dot product (and it's not obvious why you would), you also get the wrong answer of course. It's only if you understand the result in terms of the energy in the direction of the filter vs the energy perpendicular to it that you can explain cos^2. That's easy enough to accept, I think, but the implications are pretty strange. And of course lots of stuff in QM work the same way for the same reason, with the same strange implications about determinism and hidden variables.
In order to have any rational approach to living in the universe, you have to assume some fundamental stuff - basically that induction works (sense data may be inaccurate, and occasionally way off, but there's some objective universe and some validity in our sense of it). If you accept that axiom, then science works and there's no mystery at all to this - you're tricking your brain into believing that random crap is profound, full stop.
If instead you reject that axiom, then there's no basis for rationality, but that being the case we're free to choose any system of reasoning that pleases us, and I choose the one that works in a rational universe, since it's as good as any.
it "should" only slow down if everyone is using it at the same time and some of those people are further away then you are to their wireless router.
That's pretty much the case with 15' wide, 40' long townhouses all in a row, where everyone has modern WiFi routers (and there's also a row of storefronts along the back of the townhouses, and some businesses there offering free WiFi to customers, to further crowd the area). I just use long ethernet cables for the important endpoints.
Some engines don't give you much choice. I'd hope modern games aren't still stuck in the single-threaded, hard-clocked world of yesteryear, but you never know.
It's also possible they're using some very slow high-level language for the AI, and/or that no one's ever done an algorithmic optimization pass on the AI code, and they just couldn't keep up with the pipeline of collision events and whatnot that are often tightly coupled with framerate.
I've been amazed in some MMOs how server performance will be totally trashed by some patch to the AI, and completely restored by the next. Poorly thought out AI code can certainly bring a CPU to its knees.
Slashdot Mirror
It's hard to google pre-heartbleed OpenSSL flaws, but there were some serious, subtle flaws in OpenSSL that looked remarkably like typos. After the NSA leaks, there's no doubt: someone committed those flaws deliberately. And the NSA leaks showed a large and well-funded program to do just that: to subvert every public cryptographic tool and standard in subtle ways, vulnerabilities that left tools secure unless you knew about the backdoor (which is particularly pernicious, as when the backdoor is inevitably discovered, the tools are in widespread use).
The open/closed source debate is like a school yard brawl when the Marines land - entirely trumped by vastly more resources spent subverting the tools than went in to writing them. What a damn waste.
With open source, you can start making your own version and modifications
That is the one real advantage. It's not cheap or easy. It's not going to be a hobby project. But it's possible.
The replacement of OpenSSL, the TrueCrypt audit and fork. That's where you see open source step ahead.
We already have countless pieces of evidence of companies being in bed with the government
There's a big difference between a company giving data to the government -- security doesn't enter into that -- and adding deliberate flaws to security products. There hasn't been much evidence of the latter, though wasn't RSA tainted? The bigger worry with proprietary security products is that they're scams, and that happens a lot, but that's a different issue.
Are you looking at the code? I don't think that's relevant.
Companies like Google, Apple, and yes Microsoft have plenty of smart people looking at their closed code for security flaws - well-trained people who's day job is to do just that.
The once-believed advantage of open-source was that companies might be in bed with the NSA, putting flaws in deliberately, but open-source projects wouldn't be. Turns out, not so much. Both groups are just as vulnerable to malicious insiders, and both are filled with techies who would be quite angry to discover a flaw deliberately hidden in their codebase.
You seem to be confusing "tech worker" and "software developer". Be the latter. Even as a software developer, first-job pay will be crap, but who cares. Once you demonstrate you can actually code professionally, it's among the best-paying jobs in America.
But you have to be able to code.
It's not about the obvious backdoor. It's often about the random number generator used for generating keys. Maybe that keyspace is smaller than you think.
How many of the e.g.cyanogenmod people collect a paycheck from the NSA? We've seen very subtle flaws in open source code that looked plausibly like a typo, but weakens security just enough for a powerful attacker while remaining secure from a script kiddy.
Not like it's just open source. Trust was lost for the hardware RNG in Intel CPUs (I'm not sure there was ever any evidence of tampering, only evidence of how subtly it could be done: only one guy messing with a mask at the last minute, and the RNG output would still look random).
Well, good luck to you, but whatever you're doing doesn't seem to be making you happy. For sure, if you believe there's no winning strategy, you're not going to find one. Don't talk yourself into hopelessness.
Are you seriously surrounded by people who mow their own yards? Can you even buy a car with no airbags? Are you tired of unlimited free porn? Do you mistake a down business cycle for a non-cyclic trend?
The world didn't end after the 30s, or the 70s, and it won't end after the past 10 years either. Whatever life choices you've made that have left you so pessimistic and unhappy (and I'm guessing unsuccessful?), maybe try different choices?
What, a time before Crocknight lied every night about the Vietnam War because he wanted us out of it? Before the a CIA agent who wanted Nixon out used a of couple newspaper reporters as tools to accomplish his political agenda? Before 5 steady years of wartime propaganda to keep morale up on the home front? Before the press conspired to hide the fact the president was a Polio victim who couldn't walk? Was there ever actually a golden age when the press wasn't just politics? I doubt it. Oh, the political agenda changes from generation to generation, but that's about it.
The only part of the paper you can believe is the sports section. I doubt it's ever been different.
But you can declare, in your docs, what accuracy you promise, and then either succeed or fail at that. And that's the issue here, no? Intel fails at their own published accuracy.
The positive or the negative root?
I just use the average of the two, for predictable output.
How is there no "correct answer" to e.g cos(0.4) to some specified accuracy? It's right or it's not. Sure, there may be some undefined bits if the precision of the result exceeds the required accuracy, but to whatever accuracy was promised, the result is correct or not.
Sony? Really? The Justice Department bitchslapped Sony so hard the shareholders cried (they missed their quarter, by a lot, due to the fines and the CEO resigned) over the rootkit fiasco. There are real examples of corporate corruption of the justice system, but that's not one of them.
That reasoning is too specific to this one experiment - you always get that cos^2 out of the QM for cases like this. And the chance that a single photon passes through really does go by cos^2 - that's the lesson here, IMO, that energy isn't just an accident that we happen to be squaring something before we measure, but instead that it's the common pattern we observe across both QM and relativity. There's always some sort of vector projection and then squaring going on, and while what the vector is/measures changes from system to system, the square of it keeps looking like some kind of energy.
Heck, given how well Hamiltonians and Lagrangians work in explaining mechanics, I'm starting to think of force as an accidental by-product of potential energy changing over distance (which again makes sense across both QM and GR).
But then that's the real lesson here: that what our intuitions tell us are the fundamental parts of physics from which other stuff should be deduced all comes from human-scale interaction. Pretty silly to think that would be right, given that's not at all the scale at which the fundamental interactions are happening.
Well, I entire disagree with the data your reasoning from - this idea that somehow things are worse - so I doubt we'll agree on anything else.
You do some optimization on code as you write, but with AI code you can miss big, leave n^2 stuff in that your abstraction layers hid from you. It's not like you were ignoring performance when your wrote it, but when you're trying to get the functionality write you're looking at a tall pile of abstraction.
Once it works, there's often huge wins available from care refactoring for performance. But it's shortchanging the guy who wrote it to say "no one's ever optimized it".
Ow, my brain! Please, oh please, could someone replace this account with a posting robot?
For angular difference, consider the following:
You have a baseball bat at a random angle in the same plane as a picket fence. The spacing on the pickets is such that the bat will fit through if it's within 45% of vertical (or whatever direction the pickets go), and won't fit through at a greater angle. If the distribution of bat angles is even, what's the odds that a bat that passes through the first fence pass though a second at an angle of theta to the first?
It falls off directly with the angular difference. At a theta of 45 degrees, half of possible bats that fit through the first also fit through the second. At a theta of 60 degrees, one third of bats that fit through the first fit through the second. It's the same as taking two quarter-circles, and finding the overlapping length of arc when they're offset by theta (area/length of arc is where "proportional to the angular difference" would come from), as each unit-length of arc represents an equal change of encountering a bat at that range of angles.
That's the wrong answer, of course. Also, any model by which the first fence modifies the photon-substitute in some way gets very odd for the case of entangled pairs of photons, as then you need spooky action at a distance (so that modifying one instantly modifies the other, making relativity cry), and could easily get the wrong answer for 3 filters in a row.
Vector projection would give cos(theta), of course, which is also wrong, unless I misunderstand what you're saying. It's only the square of the vector projection, an energy measurement, that gets you there. I've never met anyone who would have guessed cos^2 ahead of time, but it doesn't really seem that strange that it should be energy that matters once you figure it out - why not energy, after all? (And the energy metaphor works well later on, if you accept the chance that a particle might exist in a certain place as a form of potential energy, in an energy-mass equivalence sort of way).
Anyhow, all of this seems very teachable to high-school students, and simple experiments should establish that light behaves neither light a wave nor like a particle, but "behaves in its own inimical way".
If science works, then we can accept scientific research that altering brain chemistry has a powerful affect on our ability to think rationally, and not for the better. There's no ESP to activate, no magical organ to access, there's just misbehaving wetware if the chemistry gets screwed up (or, if you have evidence otherwise, James Randi has $1 M for you).
Judgment is important. The sense of what's important and what's not is, after all, going to influence your success in life greatly, however you might measure that. If you're acting effectively at random, you're unlikely to get vey far, and focusing on the wrong things to spend your mental effort on is nearly as bad.
It's possible to imagine a drug that would give you deep insight into reality, not mere entertainment. But defeating your judgment such that you mistaken assign deep meaning to random sensation is not that, however entertaining it might be.
True enough, but it's a compelling illusion. Stuff that would dismissed after a college late-night rambling discussion of everything gets instead taken as a profound insight into the workings of the universe. But the drug didn't give you any more insight than you had before, it merely made you think your conclusion was more weighty than it was.
Two-earner households aren't that old. While women went to work during WWII, they stopped when the men came back home.
On the other hand, the automation that the Luddites fought against caused the better part of a century's worth of economic hardship, but it did eventually recover, so maybe it is just a dip long enough to last generations-worth of working lives.
You have an odd view of history if you think the industrial revolution was a bad thing for our standard of living, but one way or another this will be the same sort. I think you seriously underestimate the benefit of technology on our daily lives (technology: the ability to produce a good or deliver a service more efficiently - consuming less labor, or less power, or less raw materials, or all of these). It's what enables exponential growth of the economy at constant population.
After all, the average standard of living is simply all the goods and services all of us produce, divided by the population. Technology is really the only way to drive that average up, if you think about it. And while "net worth" may have very unequal distribution, goods and services are remarkably equal these days, at least in the West, if you ignore social status markers in the mix. Stuff made by robots isn't just going to sit in a warehouse unused - one way or another, there will be that much more goods and services distributed among all of us, and exponential growth stomps minor variations in distribution of stuff, given time.
It's not vector projection in any straightforward way. With the most obvious mental model of polarization (light pointed more in the direction of the filter than perpendicular to it gets through), you expect the energy passing through both filters to fall off directly proportional to the angle between them. If you just took the dot product (and it's not obvious why you would), you also get the wrong answer of course. It's only if you understand the result in terms of the energy in the direction of the filter vs the energy perpendicular to it that you can explain cos^2. That's easy enough to accept, I think, but the implications are pretty strange. And of course lots of stuff in QM work the same way for the same reason, with the same strange implications about determinism and hidden variables.
In order to have any rational approach to living in the universe, you have to assume some fundamental stuff - basically that induction works (sense data may be inaccurate, and occasionally way off, but there's some objective universe and some validity in our sense of it). If you accept that axiom, then science works and there's no mystery at all to this - you're tricking your brain into believing that random crap is profound, full stop.
If instead you reject that axiom, then there's no basis for rationality, but that being the case we're free to choose any system of reasoning that pleases us, and I choose the one that works in a rational universe, since it's as good as any.
it "should" only slow down if everyone is using it at the same time and some of those people are further away then you are to their wireless router.
That's pretty much the case with 15' wide, 40' long townhouses all in a row, where everyone has modern WiFi routers (and there's also a row of storefronts along the back of the townhouses, and some businesses there offering free WiFi to customers, to further crowd the area). I just use long ethernet cables for the important endpoints.
Some engines don't give you much choice. I'd hope modern games aren't still stuck in the single-threaded, hard-clocked world of yesteryear, but you never know.
It's also possible they're using some very slow high-level language for the AI, and/or that no one's ever done an algorithmic optimization pass on the AI code, and they just couldn't keep up with the pipeline of collision events and whatnot that are often tightly coupled with framerate.
I've been amazed in some MMOs how server performance will be totally trashed by some patch to the AI, and completely restored by the next. Poorly thought out AI code can certainly bring a CPU to its knees.