The more that trolls gouge Apple, MS, Oracle, IBM, Google, and the rest of the big tech companies, cost them millions in legal fees, court costs, awards of damages, injunctions that compel them to pass up opportunities, inconvenience and lose customers, the more these politically weighty companies might decide to instruct their bought congress critters to consider eliminating software patents.
Nope, they'll just play the same game themselves. All of those big companies are stockpiling huge amounts of patents for the most mundane ideas, just to use as ammo against others. They don't mind paying the odd couple of million dollars every now and then. It's only the little companies and individuals who really get hurt. RIM being a notable exception, but things rarely get that bad.
It's amazing what some of these people come up with to hide malicious code using seemingly honest coding mistakes that are hard to spot. And I'm sure the NSA can do even better than them. Certainly in a huge, complex piece of code like the Linux kernel. And how many people really inspect that code anyway?
Then again, the back door would be easier to find by criminals. I don't personally care that much about the NSA snooping through my e-mails. But if some criminal can read them just as easily, it's a different story.
And then you end up with an increased insurance premium because the other driver is not involved in your accident anymore and didn't officially cause it.
Yep, the automatic car can make a perfect swerve, using exactly the amount of available space to avoid the baby by a few centimeters while the human driver will jerk the wheel, skid left, right, left, right in increasing amplitudes and veer off the road. I've actually seen something like this happen in front of me.
If these car autopilots are anything like airplane autopilots, they WILL just kick out of auto drive. I sure hope they can do better, but there will always be rare malfunctions that result in sudden loss of the autopilot.
Of course, cars do have the advantage that they can just stop where they are.
In a plane, too, the pilots need to be ready to take over on the fly all the time with little thinking time to work out why the system kicked out of autopilot mode. Especially during automatic landings. We still use autopilot for most of the flight, though, it makes our life a lot easier and gives us time to focus on other tasks, like getting weather reports and things like that.
When encountering warning signs, the car will probably just beep and let you take over control (of stop if you don't). Autodrive will still be available for the vast majority of your trip.
The car can use radar frequencies that can see right through the white wash, so it may actually do a better job than you under those conditions. And autonomous racing cars have been demonstrated skidding through turns on mountainous terrain in perfectly controlled sideslips. Avoiding potholes doesn't seem like an insurmountable task either. Of course, when the road has pretty much disappeared under a thick layer of snow, it may become more difficult but even then, the car might be able to use its database and a few cues from surrounding houses and light poles to stay on the road while you're desperately looking for some indication of where the actual road is.
In fact, your examples are actually situations where I would expect the car to do better, not worse. I'm more worried about judgement errors. How will the car navigate through really dense traffic and crossing pedestrians, for example? Current systems will probably just stop and wait for the road to be clear again. Good luck getting through an old European city center that way. And if a ball flies across the street, will the car be intelligent enough to know that a child may be running behind it? Human common sense is hard to replace.
But anyway, for all those situations, manual control is still available. I fully expect the system to work quite well on highways, which is probably what Elon Musk is aiming for right now.
Nope, even a 26 year old A320 is capable of autoland if the required ILS system is available for that runway. In low visibility conditions, we are even required to use it. But since it's actually more work than a manual landing (and requires special training), most pilots prefer manual landings if visibility is sufficient.
Well, "routinely" only in low visibility conditions. When visibility is sufficient (more than a kilometer or so), most pilots prefer to land manually.
Also, the airport has to be equipped with a special kind of Instrument Landing System with multiple failsafes and restricted areas to guard against interference. Most smaller airports just have a regular ILS not designed for autoland (and of course, many don't even have that).
But when all the required equipment is available and weather conditions dictate it, and if the crew has had the required training, then yes, most relatively recent airliners are capable of autoland.
NO, for crying out loud, you don't slow down to merge behind traffic. That disrupts traffic flow behind you and may even create traffic jams. You plan ahead, identify a gap, adjust your speed (either slightly higher or lower, depending on traffic ahead of you, but preferably higher), and move into the gap well before the exit.
It's quite infuriating to see people braking, and making all the cars behind them brake as well, to squeeze into a short gap behind the car next to them while there's a huge gap in front of that car that they can effortlessly get into without hindering anybody, just by increasing their speed a little bit. Slower is not always better!
Of course that doesn't mean you should floor it, pass as many cars as you can before the exit, then throw yourself into the last possible gap while slamming the brakes. Like I said, plan ahead. Just don't default to slowing down as the only viable option. Be flexible and reasonable.
Basically, there's no such thing as "space itself", there's no "aether" that carries light, that's the whole point of relativity. Only the relative motions between objects are relevant. And distant objects are really moving kinetically away from us.
However, if we choose a coordinate system that is tailor-made to our expanding universe, stretched and sped up to undo any dilations caused by the expansion of the universe, motion does appear to be relative to some kind of backgrond structure that is expanding faster than the speed of light. This is a mathematical result of our choice of coordinates, it's not a physically "real" thing. It does make life a bit easier for cosmologists and is used so often that they don't even bother to mention it, which is the cause for a lot of confusion.
Really, click the link to find out more, it got a +5 rating and quite a few replies thanking me for the explanation.
Basically, it all depends on how you define distances and times. With a metric that obeys Special Relativity, nothing is faster than the speed of light but the universe is not homogenous because of Lorentz contraction and time dilation. It looks really weird and subjective with us at the center being the only "normal" part of the universe. You can fix that by using comoving coordinates, defining distances and times differently. This way, you get rid of the relativistic distortions due to the expansion speed of the universe, making everything nicely homogenous, but you give up the constancy of the speed of light. Speed of light is now relative to "space itself" and space expands faster than light. Objects that will never be visible to us using the second metric, will never even exist in the first metric since they are infinitely far in the future (their passage of time being slowed to an asymptotic halt by time dilation)
Pick your metric, both are equally valid, but most people seem to prefer comoving coordinates.
Reversible ciphers by definition don't decrease entropy. Hash functions do.
Oops, of course.
You'd have to show that, especially for the case when one of the sources is fully compromised. (I don't believe RdRand is, just saying it's not that simple...)
If you have a relatively good random stream, and you xor it with the fully compromised stream, the best the NSA can do is xor everything with their known stream and arrive back at your original random stream. How would that ever be worse? And you will still have improved security against non-NSA eavesdroppers. The only bad scenario I can think of, is that the NSA would target one specific algorithm, in a known binary, and use the data from certain registers that contain information about the other random number generator or the original plaintext in such a way that the combination of the random numbers is no longer random. But that would be pretty far-fetched imho.
If you have a perfectly random source, and you xor it with anything, the result will still be perfectly random. Pretty much by definition.
I'm not an expert on cryptography, but I would guess that if you have a pretty good, not quite perfect random generator and you xor it with some other, *unrelated* random generator, the resulting random numbers should be better, not worse. So if you get data from disk drive response times and stuff like that to make something that's reasonably random, and then xor that with the NSA random stream, I would wager that this could only make your random data better, not worse.
Scrambling random data with ciphers or hash functions before using it is a different matter, since those actually decrease entropy. Also, xor'ing two pseudo-random streams from similar algorithms is probably not a good idea since any similarities might xor into predictable outcomes. But as long as the two streams are from a sufficiently different source, things can only get better by xoring them.
RDRAND is an instruction, just like "add these two registers" or "jump to this address". Of course it's still available to user space applications. The point is that you can specify that the OS itself should not use it for things like/dev/random. If a user space application wants to use it, there's not much the kernel can/should do about that. Apps can use pretty much any insecure random algorithm anyway.
The arms carrying the letters make a slightly different sound when hitting the paper. Extremely sensitive NSA spy satellites can pick up these sounds and eavesdrop on your entire message.
Hashes have collisions: multipe strings can result in the same hash, although hash designers try to minimise this as much as possible. If you keep hashing the password over and over again, like 5000 times, the resulting number of possible results will get smaller and smaller and therefore the final key, which you use to encrypt the actual file, will be less safe. Not a good idea.
Different compilers produce different code. Some will produce faster and/or smaller code, even with optimisations turned off. Especially (obviously) code compiled for a different CPU. There's no way you'll find a back door by simply comparing the size of the binaries. They will always be different anyway.
No, it doesn't even make it through a day, not by a long shot. I guess you need to buy two if you want to be able to know what time it is from morning till evening without taking your cell phone out of your pocket.
Slashdot Mirror
The more that trolls gouge Apple, MS, Oracle, IBM, Google, and the rest of the big tech companies, cost them millions in legal fees, court costs, awards of damages, injunctions that compel them to pass up opportunities, inconvenience and lose customers, the more these politically weighty companies might decide to instruct their bought congress critters to consider eliminating software patents.
Nope, they'll just play the same game themselves. All of those big companies are stockpiling huge amounts of patents for the most mundane ideas, just to use as ammo against others. They don't mind paying the odd couple of million dollars every now and then. It's only the little companies and individuals who really get hurt. RIM being a notable exception, but things rarely get that bad.
http://underhanded.xcott.com/
It's amazing what some of these people come up with to hide malicious code using seemingly honest coding mistakes that are hard to spot. And I'm sure the NSA can do even better than them. Certainly in a huge, complex piece of code like the Linux kernel. And how many people really inspect that code anyway?
Then again, the back door would be easier to find by criminals. I don't personally care that much about the NSA snooping through my e-mails. But if some criminal can read them just as easily, it's a different story.
And then you end up with an increased insurance premium because the other driver is not involved in your accident anymore and didn't officially cause it.
Yep, the automatic car can make a perfect swerve, using exactly the amount of available space to avoid the baby by a few centimeters while the human driver will jerk the wheel, skid left, right, left, right in increasing amplitudes and veer off the road. I've actually seen something like this happen in front of me.
Indeed, that's exactly what they would do.
Of course if they can decrease gravity, they can also increase it when required. You'll be able to corner better than a F1 car.
If these car autopilots are anything like airplane autopilots, they WILL just kick out of auto drive. I sure hope they can do better, but there will always be rare malfunctions that result in sudden loss of the autopilot.
Of course, cars do have the advantage that they can just stop where they are.
Until one accident involving a malfunction of an automatic car is shown on Fox News.
In a plane, too, the pilots need to be ready to take over on the fly all the time with little thinking time to work out why the system kicked out of autopilot mode. Especially during automatic landings. We still use autopilot for most of the flight, though, it makes our life a lot easier and gives us time to focus on other tasks, like getting weather reports and things like that.
That's a full 4%!
When encountering warning signs, the car will probably just beep and let you take over control (of stop if you don't). Autodrive will still be available for the vast majority of your trip.
The car can use radar frequencies that can see right through the white wash, so it may actually do a better job than you under those conditions. And autonomous racing cars have been demonstrated skidding through turns on mountainous terrain in perfectly controlled sideslips. Avoiding potholes doesn't seem like an insurmountable task either. Of course, when the road has pretty much disappeared under a thick layer of snow, it may become more difficult but even then, the car might be able to use its database and a few cues from surrounding houses and light poles to stay on the road while you're desperately looking for some indication of where the actual road is.
In fact, your examples are actually situations where I would expect the car to do better, not worse. I'm more worried about judgement errors. How will the car navigate through really dense traffic and crossing pedestrians, for example? Current systems will probably just stop and wait for the road to be clear again. Good luck getting through an old European city center that way. And if a ball flies across the street, will the car be intelligent enough to know that a child may be running behind it? Human common sense is hard to replace.
But anyway, for all those situations, manual control is still available. I fully expect the system to work quite well on highways, which is probably what Elon Musk is aiming for right now.
Nope, even a 26 year old A320 is capable of autoland if the required ILS system is available for that runway. In low visibility conditions, we are even required to use it. But since it's actually more work than a manual landing (and requires special training), most pilots prefer manual landings if visibility is sufficient.
Well, "routinely" only in low visibility conditions. When visibility is sufficient (more than a kilometer or so), most pilots prefer to land manually.
Also, the airport has to be equipped with a special kind of Instrument Landing System with multiple failsafes and restricted areas to guard against interference. Most smaller airports just have a regular ILS not designed for autoland (and of course, many don't even have that).
But when all the required equipment is available and weather conditions dictate it, and if the crew has had the required training, then yes, most relatively recent airliners are capable of autoland.
NO, for crying out loud, you don't slow down to merge behind traffic. That disrupts traffic flow behind you and may even create traffic jams. You plan ahead, identify a gap, adjust your speed (either slightly higher or lower, depending on traffic ahead of you, but preferably higher), and move into the gap well before the exit.
It's quite infuriating to see people braking, and making all the cars behind them brake as well, to squeeze into a short gap behind the car next to them while there's a huge gap in front of that car that they can effortlessly get into without hindering anybody, just by increasing their speed a little bit. Slower is not always better!
Of course that doesn't mean you should floor it, pass as many cars as you can before the exit, then throw yourself into the last possible gap while slamming the brakes. Like I said, plan ahead. Just don't default to slowing down as the only viable option. Be flexible and reasonable.
I don't think you read the link I posted, I will repeat it here:
http://hardware.slashdot.org/comments.pl?sid=3900797&cid=44098889
Basically, there's no such thing as "space itself", there's no "aether" that carries light, that's the whole point of relativity. Only the relative motions between objects are relevant. And distant objects are really moving kinetically away from us.
However, if we choose a coordinate system that is tailor-made to our expanding universe, stretched and sped up to undo any dilations caused by the expansion of the universe, motion does appear to be relative to some kind of backgrond structure that is expanding faster than the speed of light. This is a mathematical result of our choice of coordinates, it's not a physically "real" thing. It does make life a bit easier for cosmologists and is used so often that they don't even bother to mention it, which is the cause for a lot of confusion.
Really, click the link to find out more, it got a +5 rating and quite a few replies thanking me for the explanation.
I wrote a mini-article about "space itself expanding faster than light" in a Slashdot comment some time ago:
http://hardware.slashdot.org/comments.pl?sid=3900797&cid=44098889
Basically, it all depends on how you define distances and times. With a metric that obeys Special Relativity, nothing is faster than the speed of light but the universe is not homogenous because of Lorentz contraction and time dilation. It looks really weird and subjective with us at the center being the only "normal" part of the universe. You can fix that by using comoving coordinates, defining distances and times differently. This way, you get rid of the relativistic distortions due to the expansion speed of the universe, making everything nicely homogenous, but you give up the constancy of the speed of light. Speed of light is now relative to "space itself" and space expands faster than light. Objects that will never be visible to us using the second metric, will never even exist in the first metric since they are infinitely far in the future (their passage of time being slowed to an asymptotic halt by time dilation)
Pick your metric, both are equally valid, but most people seem to prefer comoving coordinates.
Reversible ciphers by definition don't decrease entropy. Hash functions do.
Oops, of course.
You'd have to show that, especially for the case when one of the sources is fully compromised. (I don't believe RdRand is, just saying it's not that simple...)
If you have a relatively good random stream, and you xor it with the fully compromised stream, the best the NSA can do is xor everything with their known stream and arrive back at your original random stream. How would that ever be worse? And you will still have improved security against non-NSA eavesdroppers.
The only bad scenario I can think of, is that the NSA would target one specific algorithm, in a known binary, and use the data from certain registers that contain information about the other random number generator or the original plaintext in such a way that the combination of the random numbers is no longer random. But that would be pretty far-fetched imho.
If you have a perfectly random source, and you xor it with anything, the result will still be perfectly random. Pretty much by definition.
I'm not an expert on cryptography, but I would guess that if you have a pretty good, not quite perfect random generator and you xor it with some other, *unrelated* random generator, the resulting random numbers should be better, not worse. So if you get data from disk drive response times and stuff like that to make something that's reasonably random, and then xor that with the NSA random stream, I would wager that this could only make your random data better, not worse.
Scrambling random data with ciphers or hash functions before using it is a different matter, since those actually decrease entropy. Also, xor'ing two pseudo-random streams from similar algorithms is probably not a good idea since any similarities might xor into predictable outcomes. But as long as the two streams are from a sufficiently different source, things can only get better by xoring them.
RDRAND is an instruction, just like "add these two registers" or "jump to this address". Of course it's still available to user space applications. The point is that you can specify that the OS itself should not use it for things like /dev/random. If a user space application wants to use it, there's not much the kernel can/should do about that. Apps can use pretty much any insecure random algorithm anyway.
The arms carrying the letters make a slightly different sound when hitting the paper. Extremely sensitive NSA spy satellites can pick up these sounds and eavesdrop on your entire message.
Hashes have collisions: multipe strings can result in the same hash, although hash designers try to minimise this as much as possible. If you keep hashing the password over and over again, like 5000 times, the resulting number of possible results will get smaller and smaller and therefore the final key, which you use to encrypt the actual file, will be less safe. Not a good idea.
Different compilers produce different code. Some will produce faster and/or smaller code, even with optimisations turned off. Especially (obviously) code compiled for a different CPU. There's no way you'll find a back door by simply comparing the size of the binaries. They will always be different anyway.
No, it doesn't even make it through a day, not by a long shot. I guess you need to buy two if you want to be able to know what time it is from morning till evening without taking your cell phone out of your pocket.