Algorithms have been killing people for years.
There are anti missile systems which target according to their threat assessment.
Systems have been developed which require a dead man switch once armed, if the operator dies the system takes that as permission to kill anything that looks like a threat and can't convince it otherwise.
This horse left the stable decades ago.
Logs are a series of time stamped records and events. You have to read the logs to build the graphs.
If someone is supposed to be driving from DC to Boston on I95 and the position logs put them in Manhattan for part of their journey then any fool can tell that our journo didn't play by the rules. So now he has been caught lying, and even if lying is normal being caught is embarrassing.
But they block Americans because the universities prefer foreign students because they pay more. Neither my wife nor my daughter can get in to the courses they want in the University of California system because the courses they want are full of foreigners. I am so glad I got my free education in Newcastle upon Tyne. My university doesn't have huge stadiums or any great facilities outside of a gymnasium and a decent library, but it does have a lot of class rooms and taught a lot of knowledge to a lot of kids and almost all for free. If many of your brightest people are priced out of education where is the country going?
They do pay more. Which is why American students can't get in to their local universities. That is just wrong. Universities should be run at cost and shouldn't pay anyone over the going rate for the job. Maybe a couple of hundred thousand for the top job and that's it. Places lice UC Berkeley should have to take their Californian students in preference to foreign students. Maybe allow 15% out of state and 10% foreign but allocate 75% to local kids, unless they don't apply. Education should be the primary role of these places, not making cash for their management.
Seriously? You think Microsoft invented the mouse and keyboard? The first machine to have a mouse was the Xerox 8010 in 1981, it was integral to the Mac in 1984. Windows didn't turn up in a form that could use a mouse until the late 80s. My Counting House GT2000 in 1982 had a lovely separate keyboard, I have never found another like it. All this while MS was still just some horrible nightmare, back when the latest machines had 12.5MHz 68000s with nice orthogonal architectures and running Unix ten times faster than the first IBM PC.
So the claims they used to justify the ETOPS changes were that the chances of losing two engines was one in umpteen millions. I know there was one due to loss of oil back in about 1983 and I think there's been another since then. I don't count the Gimli Glider because that was a fuelling issue, but I'm still happier with four engines than two.
Takeoff is easy, any fool can do that, not that I'd trust me to fly a 747. Decent landings are a good deal more difficult, but airlines still don't require a PhD for their pilots. The constraints on a carrier landing are way tougher, you are actually aiming for a tiny target at high speed and the consequences of getting it wrong can be extreme. Landing a light aircraft on a commrcial runway is relatively easy, 60mph and more than a mile of concrete to choose from at Oakland for example. Of course there are more challenging grass strips, but thousands of pilots make uneventful landings every day.
The ESS escape shuttle is flown and landed by software running on top of VxWorks, I saw the press announcements a few years ago after their test flight. Orbit to airport in one easy lesson.
Garbage.That's the Mulhouse-Habsheim crash and was caused by the pilot doing an illegal (can't take fare paying passangers to an airshow as an exhibit) low and slow pass. The co-pilot actually warned him that he was too low but the pilot didn't apply power until the aircraft was already bound to crash. He didn't move the throttles until less than seven seconds from impact, on the tapes you can hear the engines just coming up from idle as the aircraft hits the trees.
Sadly TLC used this film and said it was the first example of a fully computer controlled landing, and sadly people believed them. Autopilots have been landing aircraft for decades.
Back in 1983 I was at the Royal Aircraft Establishment in Bedford and we were testing the autopilot's ability to predict ETA at 3D points. The only thing not controlled by the autopilot on XX105 in 1983 was the spoilers. It needed some work on the sharp turns and lost lock on the radio beacons on short finals, so it wasn't perfect, but it was just an experiment. It was also the first aircraft with a glass cockpit IIRC, at least on the co-pilot's side, pilots controls were conventional. It controlled the NAV radios as well because we were navigating by automatically switching the DME beacons. The previous year they had been to the Paris airshow with the same aircraft and demonstrated voice control of the comms radios, so far as I know that still hasn't reached service.
Not true for commercial aircraft though. Most of those have FMS which run the engines as well as the flight plan, even an old 747 I was on had had a decent FMS retrofitted, it's essential for fuel economy. Crappy old aircraft may not have much of an autopilot, but then they don't have any business cruising at FL410 with a hundred people on board. Sadly I expect there are still some old wrecks wandering the world.
Hmm, limited availability of autoland? I'm near OAK, SJC and SFO, they all have more than one ILS. My experience with GPS has been that it can place you within about ten feet, but GPS approaches aren't common yet. Certainly anywhere in the states you are usually close to an airfield of some sort with ILS
There are terrain avoidance systems that know every hill in the world, all you actually need is minimum safe altitude for your area.
Similarly for other aircraft, there's mode-s transponders and TCAS, you'd also expect the transponder to squawk 7700 so any ATC would know you were in trouble.
Linking to weather radar would be an issue, but if you'd diverted around weather you'd probably be on a heading hold anyway.
Yes, I do. Range goes to hell when altitude gets low, but alive and short of range is a way better situation than dead and cruising on into the wide blue yonder.
Yes, it's complex, but when the problem is "we're all dying" and answer of "do you have any idea how complex the fueling map is for this engine" isn't in any way helpful. Lots and lots of complex systems, but mostly they just keep on keeping on, their complexity is all well contained and they have a relatively simple published interface.
I've worked on avionics on and off since 1982. I've also never met a system with no bugs, but the external checks prevent the bugs in components from trashing the system, mostly. Having said that, military aircraft have been lost to software issues, the ones I know were flight control gain related though.
Oh save me from the "it's frightfully complex" brush-off. It really isn't that complex and the real problem is more crew acceptance than anything else. Many pilots routinely use autoland these days, Southwest actually had it disabled so that they could be certain the pilots would stay in practice.
Mostly when you get down to 15k people would come round, if the system was quick enough, and at that point the pilot can fix it himself. Losing altitude trashes your range, that could be a real issue.
The systems are complex, but replacing the flight plan automatically when the crew fails to respond is not an impossible task.
A Garmin aviation model GPS will tell you the nearest diversion airfield at the press of a button. X-plane is an FAA approved simulator, at least for navigation work, that can tell you the nearest airfield anywhere in the world. I'd expect a flight standard system to be engineered a lot more tightly, but it's far from impossible.
That's far from true, the French just started formal legal investigations into the retired chief engineer of the French part of the Concorde program for failing to address a known weakness in French operated Concorde aircraft which led to the fatal crash near Paris in 2000 which killed 113.
http://www.guardian.co.uk/airlines/story/0,1371,15 79800,00.html
Then there's the ongoing ungreased jack screw issues on Alaskan Airlines MD83s even though it was the causal fact in a fatal crash.
The 777 still suffered two decompressions during testing, one when a duct popped off and another when valve broke. There's plenty to go wrong with planes even when the software works fine. Don't bother reading anything more into this than I said, I think others are giving a decent airing to the facts.
Slashdot Mirror
Have you considered a small black hole?
Algorithms have been killing people for years. There are anti missile systems which target according to their threat assessment. Systems have been developed which require a dead man switch once armed, if the operator dies the system takes that as permission to kill anything that looks like a threat and can't convince it otherwise. This horse left the stable decades ago.
If we still used this iterative process the 787 would STILL be years from production.
Logs are a series of time stamped records and events. You have to read the logs to build the graphs. If someone is supposed to be driving from DC to Boston on I95 and the position logs put them in Manhattan for part of their journey then any fool can tell that our journo didn't play by the rules. So now he has been caught lying, and even if lying is normal being caught is embarrassing.
But they block Americans because the universities prefer foreign students because they pay more. Neither my wife nor my daughter can get in to the courses they want in the University of California system because the courses they want are full of foreigners. I am so glad I got my free education in Newcastle upon Tyne. My university doesn't have huge stadiums or any great facilities outside of a gymnasium and a decent library, but it does have a lot of class rooms and taught a lot of knowledge to a lot of kids and almost all for free. If many of your brightest people are priced out of education where is the country going?
They do pay more. Which is why American students can't get in to their local universities. That is just wrong. Universities should be run at cost and shouldn't pay anyone over the going rate for the job. Maybe a couple of hundred thousand for the top job and that's it. Places lice UC Berkeley should have to take their Californian students in preference to foreign students. Maybe allow 15% out of state and 10% foreign but allocate 75% to local kids, unless they don't apply. Education should be the primary role of these places, not making cash for their management.
Seriously? You think Microsoft invented the mouse and keyboard? The first machine to have a mouse was the Xerox 8010 in 1981, it was integral to the Mac in 1984. Windows didn't turn up in a form that could use a mouse until the late 80s. My Counting House GT2000 in 1982 had a lovely separate keyboard, I have never found another like it. All this while MS was still just some horrible nightmare, back when the latest machines had 12.5MHz 68000s with nice orthogonal architectures and running Unix ten times faster than the first IBM PC.
Didn't help them in the Greek crash did it?
So the claims they used to justify the ETOPS changes were that the chances of losing two engines was one in umpteen millions. I know there was one due to loss of oil back in about 1983 and I think there's been another since then. I don't count the Gimli Glider because that was a fuelling issue, but I'm still happier with four engines than two.
Takeoff is easy, any fool can do that, not that I'd trust me to fly a 747. Decent landings are a good deal more difficult, but airlines still don't require a PhD for their pilots. The constraints on a carrier landing are way tougher, you are actually aiming for a tiny target at high speed and the consequences of getting it wrong can be extreme. Landing a light aircraft on a commrcial runway is relatively easy, 60mph and more than a mile of concrete to choose from at Oakland for example. Of course there are more challenging grass strips, but thousands of pilots make uneventful landings every day.
The ESS escape shuttle is flown and landed by software running on top of VxWorks, I saw the press announcements a few years ago after their test flight. Orbit to airport in one easy lesson.
Garbage.That's the Mulhouse-Habsheim crash and was caused by the pilot doing an illegal (can't take fare paying passangers to an airshow as an exhibit) low and slow pass. The co-pilot actually warned him that he was too low but the pilot didn't apply power until the aircraft was already bound to crash. He didn't move the throttles until less than seven seconds from impact, on the tapes you can hear the engines just coming up from idle as the aircraft hits the trees.
Sadly TLC used this film and said it was the first example of a fully computer controlled landing, and sadly people believed them. Autopilots have been landing aircraft for decades.
Back in 1983 I was at the Royal Aircraft Establishment in Bedford and we were testing the autopilot's ability to predict ETA at 3D points. The only thing not controlled by the autopilot on XX105 in 1983 was the spoilers. It needed some work on the sharp turns and lost lock on the radio beacons on short finals, so it wasn't perfect, but it was just an experiment. It was also the first aircraft with a glass cockpit IIRC, at least on the co-pilot's side, pilots controls were conventional. It controlled the NAV radios as well because we were navigating by automatically switching the DME beacons. The previous year they had been to the Paris airshow with the same aircraft and demonstrated voice control of the comms radios, so far as I know that still hasn't reached service.
Not true for commercial aircraft though. Most of those have FMS which run the engines as well as the flight plan, even an old 747 I was on had had a decent FMS retrofitted, it's essential for fuel economy. Crappy old aircraft may not have much of an autopilot, but then they don't have any business cruising at FL410 with a hundred people on board. Sadly I expect there are still some old wrecks wandering the world.
Hmm, limited availability of autoland? I'm near OAK, SJC and SFO, they all have more than one ILS. My experience with GPS has been that it can place you within about ten feet, but GPS approaches aren't common yet. Certainly anywhere in the states you are usually close to an airfield of some sort with ILS
There are terrain avoidance systems that know every hill in the world, all you actually need is minimum safe altitude for your area. Similarly for other aircraft, there's mode-s transponders and TCAS, you'd also expect the transponder to squawk 7700 so any ATC would know you were in trouble. Linking to weather radar would be an issue, but if you'd diverted around weather you'd probably be on a heading hold anyway. Yes, I do. Range goes to hell when altitude gets low, but alive and short of range is a way better situation than dead and cruising on into the wide blue yonder. Yes, it's complex, but when the problem is "we're all dying" and answer of "do you have any idea how complex the fueling map is for this engine" isn't in any way helpful. Lots and lots of complex systems, but mostly they just keep on keeping on, their complexity is all well contained and they have a relatively simple published interface. I've worked on avionics on and off since 1982. I've also never met a system with no bugs, but the external checks prevent the bugs in components from trashing the system, mostly. Having said that, military aircraft have been lost to software issues, the ones I know were flight control gain related though.
Oh save me from the "it's frightfully complex" brush-off. It really isn't that complex and the real problem is more crew acceptance than anything else. Many pilots routinely use autoland these days, Southwest actually had it disabled so that they could be certain the pilots would stay in practice. Mostly when you get down to 15k people would come round, if the system was quick enough, and at that point the pilot can fix it himself. Losing altitude trashes your range, that could be a real issue. The systems are complex, but replacing the flight plan automatically when the crew fails to respond is not an impossible task. A Garmin aviation model GPS will tell you the nearest diversion airfield at the press of a button. X-plane is an FAA approved simulator, at least for navigation work, that can tell you the nearest airfield anywhere in the world. I'd expect a flight standard system to be engineered a lot more tightly, but it's far from impossible.
That's far from true, the French just started formal legal investigations into the retired chief engineer of the French part of the Concorde program for failing to address a known weakness in French operated Concorde aircraft which led to the fatal crash near Paris in 2000 which killed 113. http://www.guardian.co.uk/airlines/story/0,1371,15 79800,00.html
Then there's the ongoing ungreased jack screw issues on Alaskan Airlines MD83s even though it was the causal fact in a fatal crash.
The 777 still suffered two decompressions during testing, one when a duct popped off and another when valve broke. There's plenty to go wrong with planes even when the software works fine. Don't bother reading anything more into this than I said, I think others are giving a decent airing to the facts.