no human being has ever chosen to nose-dive under any scenario in a commercial flight. Why was an algorithm written that could do something that no one has ever wanted to do?
Actually, human pilots DO sometimes want to do this.
When you're at high altitude in a jet there's only a small window between your stall speed and your maximum speed. If your airspeed indicators ice and start giving bad data (and several other contributing machine and human failures) you can end up flying slower than you intended. The seat of the pants warnings are subtle, so it's very easy to not realize what's happening until you start to stall - or worse, until you're deeply stalled.
Airspeed would help, but jets have very long throttle lag. Thus, the solution is to nose down - hard, if you realize the circumstances are dire. Instantly this unloads the wings; very quickly gravity starts giving you airspeed; and once your engines ramp up you can level out and maintain that speed.
So the computer saw: "WTF flaking sensors... OMFG we're crazy-stalled! Nose down hard!" It was a good reaction given the perceived circumstances.
three independent airplane computer systems are written from spec by different teams
And there's your single point of failure - the spec was defective.
In this case, one of the three sensors gave bad data. The programs correctly handled this situation, but failed to handle the second round of bad data, a failure mode unanticipated by the spec. All three redundant systems then had the same response: "sensors indicate imminent stall, nose down to recover". And since all three agreed on that course of action, it nosed down.
You state that as if it were a fact, when the reality is your trying to substitute "cheaper' for 'efficient" and hope nobody notices
Cost is one half of the measurement of cost efficiency. The other half is how much work you achieve. You were quantifying it in miles covered, so why is it wrong for me to say "efficiency = cost / miles driven"?
Or once again, you're trying to move goalposts and change definitions, while hoping vainly nobody notices
Erm, are you mistaking me for someone else? I don't think we've ever met before.
Well, no, once again you're not only wrong - but stupid. We aren't sending a replacement even though it got stuck.
Fuck you too?:) I'll give you the benefit of the doubt there.
We aren't sending more because all the money got blown on wasteful programs. And no, it's not ever going to be send an exact replacement; ideally it would be running a dozen of these programs in parallel, and when one gets stuck in the mud you just move on to the next one. You don't send an exact replacement unless it was just spewing good data.
you are utterly ignorant of what was actually accomplished by either program
You're the one who started quantifying everything in miles driven.
You also fail to consider that currently, everywhere it's practical to send men rather than robots - we send men
I understand that just fine. I just think that we're sending men to do a whole lot of jobs where it ISN'T practical.
Sign me up. Hell, sign me up for a one-way trip to Mars. I'll do it today if you give me enough research gear and supplies to keep me going for a few months when I get there, and I know I'm not the only one.
Total cost of the Apollo program in 1999 dollars: $~100B. Total cost of Spirit + Opportunity: $~1B.
Apollo cost per mile: $~4B. Rover cost per mile: $~0.04B.
That doesn't even account for the fact that Mars is a MUCH more difficult place to reach.
I don't really care if my data has a few years extra latency. Those little guys are vastly more efficient, and I'd very much rather have a hundred small-scale programs like that going on than to run another Apollo-scale program. For the amount invested we'd just be streaming down GOBS of data in comparison.
Like you say there are things that humans can do better sometimes, but dragging the rover out of the ditch isn't a good example: If it was a robot, we'd just shrug and send off another one thereby increasing the cost of the mission from 1% of Apollo to 2%.
That's not to say I'm against manned spaceflight generally. I think it's important for us to keep venturing off of this rock. Short term, though, there are a lot of other things that should be done first. Longer term, let the commercial sector work on manned spaceflight for a while. They'll get the costs down in ways NASA can't dream of.
best buy was even surprised by the sound when I returned it
This, right here, is why your experience is an anecdote, not data. You're sick of this brand because of this, but this experience is an outlier in the eyes of someone who has a much larger data set.
Not all drives are created equal, but the way people form their opinions on which ones are crap are highly nonscientific.
XFCE doesn't NEED to move fast. That's the whole idea: they don't keep changing it around and adding bells and whistles every six months. If you want shiny new features, it's not for you.
I'm sure a lot of my comfort with Gnome2 was simply familiarity, not an inherent superiority. The OP was asking for a good replacement for Gnome2... And as a Gnome2 refugee, XFCE came pretty naturally to me where KDE did not.
KDE still feels overly complicated whenever I go to configure things. Too many options makes it hard to find the one I want... And this is coming from someone who likes buying things with too many knobs.
XFCE has been treating me well. You might want to give it a try.
Basically instead of letting junkies do crimes to get their hand at illegal drugs, let's doctors prescribe it, with the official "goal" of getting the junkie some time in the far future clean).
And it works! If you give a junkie a reliable, free supply of opiates, they quit the cycle of binging then stealing things when they run out, are generally able to function in society, and gradually wean themselves off. It is more effective than any other treatment.
Methadone is particularly effective because for this because it's very long-acting. It doesn't provide a reward rush when you take it, and it doesn't crash fast leaving them desperately craving.
they are forced to go in daily to the pharmacy and consume it on site.
Sure. Heroin users are used to gauging their dose by the immediate response. Methadone is really slow, so they think they didn't take enough and take more, only to end up overdosing when it hits. For non-addiction prescriptions they just give you a 30-day supply.
And using heavy addiction inducing drugs as a general pain killer medication is not sound policy
What would you suggest for severe pain? Advil isn't going to do it.
Especially since correctly timing the spoofed GPS signals requires knowing the location of the (stealth) drone you're trying to trick.
I replied to GP but just so you see it:
You don't actually have to know the drone's location accurately. You just take a good guess and start transmitting a stream of where you want it to THINK it is. It will see a small jump in position if you guess wrong, and it will see a small jump in time if you guess the range from your transmitter to the drone wrong, but unless those trigger a fallback mechanism, you're set.
Of course you have to detect that it's there and at least get a rough position from ground sightings.
But "spoofing" GPS signals is a great deal more challenging. It's not the data on the gps signal, it's the timing that is the position information.
It's actually not very hard. You just spoof the signal (several overlaid satellite signals) of where you want the receiver to think it is. There is commercial hardware that does exactly this for GPS testing at very low power levels. Just add an RF amp and a moderate-gain antenna. It doesn't take much to completely overpower the real signal.
Those signals have to be clocked correctly relative to each other, but they don't have to be clocked terribly accurately to the GPS system. If you guess the range to the receiver wrong (and end up transmitting a little early or late) the receiver will just perceive a slight jump in time. If you guess the initial position wrong it just perceives a slight jump in position.
A smart receiver could notice those jumps and fall back to inertial navigation, but we already know this wasn't well-designed... it was using civilian GPS! They should have been using the P(Y) code which is encrypted for anti-spoofing.
the only difference between the soviet union and 'the west' is that 'the west' still hasnt collapsed yet.
It will. People looking at the inner structure have known it for a while. Now the cracks are starting to show on the surface where anyone can see them. It won't be long.
If you are brave try to look past how you're going to survive the collapse: try to plan how you will help rebuild it better.
Back up and encrypt your data. Then losing your laptop is just a monetary loss.
To save the annoyance and cost of broken glass and a laptop, I always keep it in an inconspicuous (somewhat old and worn) non-laptop-looking backpack, and throw it in the trunk so it's completely out of sight. I don't leave other expensive stuff laying out in the open either... A big cup of change is just asking for it.
The idea of the wire mesh bags is the cable lock prevents them from being easily carried away, and the mesh prevents slice-and-grab attacks. Anyone with real tools isn't going to be deterred, but it's enough to stop the opportunistic thief who just has a pocketknife. Like you say, it works if you lock it to something that can't be carried away, but it's useless otherwise.
There's a slight difference: The passenger knows to shut up when you have to maneuver. Even if they don't it's socially acceptable to ignore them while you do so.
Between the inability for the caller to see what's going on, lag, the 0.75 duplex nature of cellphones (they squelch when you're not talking), and social factors, it's a bigger imposition to tell someone to wait while you merge. If it's me, I just say "hang on, merging" and proceed to take as long as I need to get through the maneuver, then resume the conversation when I'm in a lane and my task load is back down.
But that can't be extrapolated to everyone. Is a salesman willing to tell his customer to shut up for thirty seconds so he can drive? Probably not.
Modern updates greatly reduced the positive void coefficient. It used to be wildly positive (4.7), which allowed running unenriched uranium on a non-heavy-water reactor. Now it's around 0.7, which gives you a lot more room for error.
The controls are considerably upgraded: no more graphite tips on the control rods, more manual control rods, more neutron absorbers, no more safety overrides, and more.
There aren't many single points of failure, but the safety margin and redundancy is much lower than western designs. A PWR can be leaking like a sieve and still maintain adequate cooling; a RBMK can hit trouble with only a few broken pipes, and as you say, there's no way to mitigate it, since it's part of the design.
They actually do have some some containment. It's not a heavy-duty all-encompassing concrete bunker like a western reactor, but there are high pressure management channels, steam condenser pools, etc. Any routine blowout will be contained... Just don't pull a Chernobyl.:)
I'd say RBMK safety has been upgraded from "Insanely Irresponsible" to "Poor".
What I find fascinating is that all these processes happen and we don't even know it.
You could ask the guy why he hesitated in his answers and it wouldn't be "Well, my cache got wiped when my environment-mapper interrupt fired". You could probe farther, "What were you thinking about when you first walked through the door?" and you still wouldn't get anything. These processes never enter our conscious mind unless the process finds something (perhaps a bear-shaped shadow in the corner) which needs immediate attention.
Thus, psychology research requires very creative experiments and careful statistical analysis to pull the signal out of the noise. You have to create observable side effects without creating false signals. I often learn more from reading about how a good experiment was designed than I do from the results.
What we learned from Google is: when you make a mistake, quickly and quietly cover it up. Definitely don't admit that you did something wrong.
CarrierIQ's got the message and is playing it smart: divert attention by saying THEY don't give information to the FBI, when really the problem is their SOFTWARE collecting information. See? No admission of guilt. Perhaps they also pay the appropriate bribes.
Most of the older ones can be easily rooted by the usual shenanigans; then once you install Cyanogenmod it's yours for life. It's much more pleasant than Apple's obsession with keeping you locked out.
jailbreakme.com isn't "follow these instructions". If you go there on an iPhone it gives you a big friendly button labelled "Jailbreak Me". You click it. Done.
Yes, I know the Kindle one is really easy too, but the bar for "World's Simplest" is one click. That's a tough act to beat.:)
Slashdot Mirror
no human being has ever chosen to nose-dive under any scenario in a commercial flight. Why was an algorithm written that could do something that no one has ever wanted to do?
Actually, human pilots DO sometimes want to do this.
When you're at high altitude in a jet there's only a small window between your stall speed and your maximum speed. If your airspeed indicators ice and start giving bad data (and several other contributing machine and human failures) you can end up flying slower than you intended. The seat of the pants warnings are subtle, so it's very easy to not realize what's happening until you start to stall - or worse, until you're deeply stalled.
Airspeed would help, but jets have very long throttle lag. Thus, the solution is to nose down - hard, if you realize the circumstances are dire. Instantly this unloads the wings; very quickly gravity starts giving you airspeed; and once your engines ramp up you can level out and maintain that speed.
So the computer saw: "WTF flaking sensors... OMFG we're crazy-stalled! Nose down hard!" It was a good reaction given the perceived circumstances.
three independent airplane computer systems are written from spec by different teams
And there's your single point of failure - the spec was defective.
In this case, one of the three sensors gave bad data. The programs correctly handled this situation, but failed to handle the second round of bad data, a failure mode unanticipated by the spec. All three redundant systems then had the same response: "sensors indicate imminent stall, nose down to recover". And since all three agreed on that course of action, it nosed down.
You state that as if it were a fact, when the reality is your trying to substitute "cheaper' for 'efficient" and hope nobody notices
Cost is one half of the measurement of cost efficiency. The other half is how much work you achieve. You were quantifying it in miles covered, so why is it wrong for me to say "efficiency = cost / miles driven"?
Or once again, you're trying to move goalposts and change definitions, while hoping vainly nobody notices
Erm, are you mistaking me for someone else? I don't think we've ever met before.
Well, no, once again you're not only wrong - but stupid. We aren't sending a replacement even though it got stuck.
Fuck you too? :) I'll give you the benefit of the doubt there.
We aren't sending more because all the money got blown on wasteful programs. And no, it's not ever going to be send an exact replacement; ideally it would be running a dozen of these programs in parallel, and when one gets stuck in the mud you just move on to the next one. You don't send an exact replacement unless it was just spewing good data.
you are utterly ignorant of what was actually accomplished by either program
You're the one who started quantifying everything in miles driven.
You also fail to consider that currently, everywhere it's practical to send men rather than robots - we send men
I understand that just fine. I just think that we're sending men to do a whole lot of jobs where it ISN'T practical.
Sign me up. Hell, sign me up for a one-way trip to Mars. I'll do it today if you give me enough research gear and supplies to keep me going for a few months when I get there, and I know I'm not the only one.
Total cost of the Apollo program in 1999 dollars: $~100B. Total cost of Spirit + Opportunity: $~1B.
Apollo cost per mile: $~4B.
Rover cost per mile: $~0.04B.
That doesn't even account for the fact that Mars is a MUCH more difficult place to reach.
I don't really care if my data has a few years extra latency. Those little guys are vastly more efficient, and I'd very much rather have a hundred small-scale programs like that going on than to run another Apollo-scale program. For the amount invested we'd just be streaming down GOBS of data in comparison.
Like you say there are things that humans can do better sometimes, but dragging the rover out of the ditch isn't a good example: If it was a robot, we'd just shrug and send off another one thereby increasing the cost of the mission from 1% of Apollo to 2%.
That's not to say I'm against manned spaceflight generally. I think it's important for us to keep venturing off of this rock. Short term, though, there are a lot of other things that should be done first. Longer term, let the commercial sector work on manned spaceflight for a while. They'll get the costs down in ways NASA can't dream of.
best buy was even surprised by the sound when I returned it
This, right here, is why your experience is an anecdote, not data. You're sick of this brand because of this, but this experience is an outlier in the eyes of someone who has a much larger data set.
Not all drives are created equal, but the way people form their opinions on which ones are crap are highly nonscientific.
XFCE doesn't NEED to move fast. That's the whole idea: they don't keep changing it around and adding bells and whistles every six months. If you want shiny new features, it's not for you.
once it fails securely erasing the data can be an issue
That's one of many good reasons for whole-disk encryption.
I'm sure a lot of my comfort with Gnome2 was simply familiarity, not an inherent superiority. The OP was asking for a good replacement for Gnome2... And as a Gnome2 refugee, XFCE came pretty naturally to me where KDE did not.
KDE still feels overly complicated whenever I go to configure things. Too many options makes it hard to find the one I want... And this is coming from someone who likes buying things with too many knobs.
XFCE has been treating me well. You might want to give it a try.
Basically instead of letting junkies do crimes to get their hand at illegal drugs, let's doctors prescribe it, with the official "goal" of getting the junkie some time in the far future clean).
And it works! If you give a junkie a reliable, free supply of opiates, they quit the cycle of binging then stealing things when they run out, are generally able to function in society, and gradually wean themselves off. It is more effective than any other treatment.
Methadone is particularly effective because for this because it's very long-acting. It doesn't provide a reward rush when you take it, and it doesn't crash fast leaving them desperately craving.
they are forced to go in daily to the pharmacy and consume it on site.
Sure. Heroin users are used to gauging their dose by the immediate response. Methadone is really slow, so they think they didn't take enough and take more, only to end up overdosing when it hits. For non-addiction prescriptions they just give you a 30-day supply.
And using heavy addiction inducing drugs as a general pain killer medication is not sound policy
What would you suggest for severe pain? Advil isn't going to do it.
Especially since correctly timing the spoofed GPS signals requires knowing the location of the (stealth) drone you're trying to trick.
I replied to GP but just so you see it:
You don't actually have to know the drone's location accurately. You just take a good guess and start transmitting a stream of where you want it to THINK it is. It will see a small jump in position if you guess wrong, and it will see a small jump in time if you guess the range from your transmitter to the drone wrong, but unless those trigger a fallback mechanism, you're set.
Of course you have to detect that it's there and at least get a rough position from ground sightings.
But "spoofing" GPS signals is a great deal more challenging. It's not the data on the gps signal, it's the timing that is the position information.
It's actually not very hard. You just spoof the signal (several overlaid satellite signals) of where you want the receiver to think it is. There is commercial hardware that does exactly this for GPS testing at very low power levels. Just add an RF amp and a moderate-gain antenna. It doesn't take much to completely overpower the real signal.
Those signals have to be clocked correctly relative to each other, but they don't have to be clocked terribly accurately to the GPS system. If you guess the range to the receiver wrong (and end up transmitting a little early or late) the receiver will just perceive a slight jump in time. If you guess the initial position wrong it just perceives a slight jump in position.
A smart receiver could notice those jumps and fall back to inertial navigation, but we already know this wasn't well-designed... it was using civilian GPS! They should have been using the P(Y) code which is encrypted for anti-spoofing.
This is why you load it up on other people's TVs.
the only difference between the soviet union and 'the west' is that 'the west' still hasnt collapsed yet.
It will. People looking at the inner structure have known it for a while. Now the cracks are starting to show on the surface where anyone can see them. It won't be long.
If you are brave try to look past how you're going to survive the collapse: try to plan how you will help rebuild it better.
Back up and encrypt your data. Then losing your laptop is just a monetary loss.
To save the annoyance and cost of broken glass and a laptop, I always keep it in an inconspicuous (somewhat old and worn) non-laptop-looking backpack, and throw it in the trunk so it's completely out of sight. I don't leave other expensive stuff laying out in the open either... A big cup of change is just asking for it.
The idea of the wire mesh bags is the cable lock prevents them from being easily carried away, and the mesh prevents slice-and-grab attacks. Anyone with real tools isn't going to be deterred, but it's enough to stop the opportunistic thief who just has a pocketknife. Like you say, it works if you lock it to something that can't be carried away, but it's useless otherwise.
There's a slight difference: The passenger knows to shut up when you have to maneuver. Even if they don't it's socially acceptable to ignore them while you do so.
Between the inability for the caller to see what's going on, lag, the 0.75 duplex nature of cellphones (they squelch when you're not talking), and social factors, it's a bigger imposition to tell someone to wait while you merge. If it's me, I just say "hang on, merging" and proceed to take as long as I need to get through the maneuver, then resume the conversation when I'm in a lane and my task load is back down.
But that can't be extrapolated to everyone. Is a salesman willing to tell his customer to shut up for thirty seconds so he can drive? Probably not.
Modern updates greatly reduced the positive void coefficient. It used to be wildly positive (4.7), which allowed running unenriched uranium on a non-heavy-water reactor. Now it's around 0.7, which gives you a lot more room for error.
The controls are considerably upgraded: no more graphite tips on the control rods, more manual control rods, more neutron absorbers, no more safety overrides, and more.
There aren't many single points of failure, but the safety margin and redundancy is much lower than western designs. A PWR can be leaking like a sieve and still maintain adequate cooling; a RBMK can hit trouble with only a few broken pipes, and as you say, there's no way to mitigate it, since it's part of the design.
They actually do have some some containment. It's not a heavy-duty all-encompassing concrete bunker like a western reactor, but there are high pressure management channels, steam condenser pools, etc. Any routine blowout will be contained... Just don't pull a Chernobyl. :)
I'd say RBMK safety has been upgraded from "Insanely Irresponsible" to "Poor".
What I find fascinating is that all these processes happen and we don't even know it.
You could ask the guy why he hesitated in his answers and it wouldn't be "Well, my cache got wiped when my environment-mapper interrupt fired". You could probe farther, "What were you thinking about when you first walked through the door?" and you still wouldn't get anything. These processes never enter our conscious mind unless the process finds something (perhaps a bear-shaped shadow in the corner) which needs immediate attention.
Thus, psychology research requires very creative experiments and careful statistical analysis to pull the signal out of the noise. You have to create observable side effects without creating false signals. I often learn more from reading about how a good experiment was designed than I do from the results.
What we learned from Google is: when you make a mistake, quickly and quietly cover it up. Definitely don't admit that you did something wrong.
CarrierIQ's got the message and is playing it smart: divert attention by saying THEY don't give information to the FBI, when really the problem is their SOFTWARE collecting information. See? No admission of guilt. Perhaps they also pay the appropriate bribes.
http://htcdev.com/bootloader/
You don't understand. It's a different crime when it has something to do with computers. That's why we need all these computer crime laws!
It's not simple for the end user, but it's officially supported on their new phones: http://htcdev.com/bootloader/
Most of the older ones can be easily rooted by the usual shenanigans; then once you install Cyanogenmod it's yours for life. It's much more pleasant than Apple's obsession with keeping you locked out.
Yes, I actually did RTFA.
jailbreakme.com isn't "follow these instructions". If you go there on an iPhone it gives you a big friendly button labelled "Jailbreak Me". You click it. Done.
Yes, I know the Kindle one is really easy too, but the bar for "World's Simplest" is one click. That's a tough act to beat. :)
When the GP said "1984 style", they were referring to the fact that Amazon actually revoked some copies of 1984 in a flash of brilliant irony.