I was involved in a similar, but very much smaller scale, experiment for my MSc thesis (JANZOS), attempting to find detect gamma rays from the (then very recent) supernova 1987A in the Large Magellanic Cloud.
So supernovae were a prime suspect source back then.
We had three (not four) 2 metre (not 12 metre) telescopes with about 30 'pixels' each (compared to a few thousand for HESS.) (I actually worked on another part of the experiment, which used particle detectors to detect higher energy showers.)
A significant problem is to distinguish between showers created by gamma rays and ones created by charged particles (mostly protons.) The charged particle showers are 'uninteresting', because the direction they come from is uncorrelated to their source - they move on curved paths due to galactic magnetic fields. Unfortunately, they are about 99% of the cosmic rays. We were not able to distinguish, so we had a large 'signal to noise' problem.
There was a single telescope similar to these ones in the mid 80s (the Whipple Telescope, I think) which claimed to be able to distinguish by details of shower structure. (We didn't have the resolution, nor perhaps the light gathering power, to make use of this.) I presume HESS has built on this work.
Note that this result does not necessarily tell us about the very highest energy cosmic rays. There is a change in the slope of the spectrum at (from memory) about 10^15 electron volts, so it is likely that different processess are involved on either side of this boundary. I think there were also theoretical reasons to think that supernovae could not accelerate particles to such high energies.
As I recall, the models for acceleration generally required shock waves in a gas with magnetic fields. Particles could repeatedly bounce across the shock, getting accelerated each time. (Think of a ball bouncing between two walls that are moving towards each other.)
Everyone I talk to seems to be "I HATE Kerry", or "I HATE Bush", but no one actually *likes* anyone anymore it seems...
Vote Cthulu 2004: When you're tired of voting for the lesser of two evils.
More seriously, the US presidential election is pretty much the ideal case for a single-transferable-vote electoral system. Just one winner, large incentive for people to study the candidates long enough to make an informed choice.
The horizon limits you to viewing half the sky. Atmospheric effects make it difficult/unviable to view close to the horizon, so in practice this is even more limited - say 1/3 of the sky. In addition, daylight restricts your observing time by a factor of more than two, and for faint diffuse objects (glaxies, nebulae) you also can't observe when a bright moon is in the sky, nor, of course, when it is cloudy - so maybe you end up with 4 hours per day of good observing time per night, on average. A space telescope suffers none of these limitations. (Well, just a little - you can't observe too close to the sun, moon or earth.)
However, although you can only observe 1/3 of the sky at a given moment, the motion of the stars through the night and year means you can observe much more of the sky if you're prepared to wait. Furthermore, if the telescope costs a small fraction of the cost of a space telescope, you can build many of them in different parts of the world, to overcome these limitations.
There are other reasons for going into space - atmospheric bluring, absorption and emission.
No - there is lots that Hubble can do that this telescope can't, and lots that no earthbound telescope can do. (See my other comment on this article: #10553382
It is another question of whether that extra capability is worth the extra cost.
The story gives the impression that the LBT will completely replace Hubble, and do a better job, while being vastly cheaper.
This is an overstatement. There is lots that Hubble can do that no other telescope can, being a unique combination of aperature (light gathering power and resolution), instruments (many wavelengths, imaging and spectroscopic) and being above the atmosphere (no 'seeing', no atmospheric absorption or emission in UV and IR.)
(This is not to downplay the LBT - doing better than HST in some aspects, and as well but much cheaper in others, is very valuable.)
Having quickly scanned the website for this telescope, I can't see how they are counteracting the bluring of 'seeing' (atmospheric turbulance). It is inconceivable that they have neglected it, but I don't see where. Adaptive optics can help, but have limitations of their own.
Another limitation of the LBT is that the high resolution reconstruction will require 3 observations at different times - so it only works well with non-time-varying targets. This is a minor limitation, however - a large majority of targets for which you want high resolution are non-variable.
Astronomers have occasionally used balloon borne telescopes for getting above most of the atmosphere, as it is much cheaper than a satellite. If there is a mass-produced long duration stratespheric balloon/airship available, it could make this much more viable.
As an aside - the article also discusses "Terahertz imaging." One terahertz corresponds to wavelength of about 0.3 mm or 300 microns - extreme IR, or short sub-millimetre, depending on your point of view.
Pedantic quibble time: Actually it is >1400 times as powerful as fusion, and even more compared to fission.
The nuclear reaction that releases the greatest percentage of mass into energy is 4 protons (hydrogen nuclei) -> 1 alpha particle (helium nucleus.) This has an 'efficiency' of about 0.7%.
This is also the reaction that powers the sun. (Technically, it is a series of reactions - either the 'p-p chain' or the 'CNO cycle', but that is too much detail to go into here.)
In principle, the best you could do would be 56 protons -> iron nucleus, but this wouldn't be feasible.
To be fair, so far as I know we haven't every built a facility dedicated to efficient large-scale production and collection of anti-matter. It wouldn't surprise me if such a facility reduced the cost by a factor of 100 or 1000. I expect that a rough design for such a facility is a goal of this research.
If we assume that the maximum cost the military would pay for enough antimater to blow up a building was $1 million, the price has to come down by a factor of 3,000,000. I really doubt this is feasible in the forseeable future (say 50 years.)
No, no, it really is a Turin machine. This is a device with internal states and a shroud that it moves back and forth, writing and removing mysterious negative images of dead people.
Don't panic/celebrate in anticipation of antimatter weapons being deployed 15 years from now.
From the article: "about 50-millionths of a gram could generate a blast equal to the explosion (roughly 4,000 pounds of TNT, according to the FBI) at the Alfred P. Murrah Federal Building in Oklahoma City in 1995."
and
"With present techniques, the price tag for 100-billionths of a gram of antimatter would be $6 billion"
from which we can calculate that blowing up a building with antimatter will cost about 3 trillion dollars. (And tens or hundreds of millions for the equipment to confine the antimatter until you want it to detonate, but that is negligible in comparison.)
Also notice that while the antimatter may be the ultimately compact explosive, the containment equipment required will increase the size of a warhead by many orders of magnitude. No antimatter rifle bullets anytime soon.
Assuming all this is so, I think a final step is required in the chain. The existence of people who can legally distribute Connolly's code does not imply that Sakic has a legally distributable copy.
Sakic needs to get a copy of his code from (directly or indirectly) one of the 'registered users', who, via the GPL, have the right to redistribute it.
This may require some persuading and legal indemnification, as the 'registered users' know that Connolly is litigation-happy and may come after them.
Possibly Sakic would need to resubmit his controversial Mambo contribution after receiving this code - ask a lawyer.
Off course, all this is moot if the Mambo code is not derivative of the Connolly code.
Alice is a contractor. She modifies a GPL program under contract for BigCorp. She sends a copy of the modifications to BigCorp.
Alice has distributed the code to BigCorp, so she has no legal right to complain should BigCorp distribute it further, so long as BigCorp do so under the GPL.
Now say Alice's contract says that she will not distribute her modifications to anyone else. Is this legal/enforcable? Did Alice have the right to modify the code while bound by such a restriction?
"Copyright infringement is a broader term and would seem to be a better fit for this situation; after all, the contract called for all copyrights to be assigned to Furthermore, Inc. But the code was a derivative of GPL-licensed code, thereby making it a derivative work of the parent code and automatically licensing it under the GNU General Public License."
But doesn't the GPL only take effect if the code is distributed? Connolly employed someone to modify GPLed code, for his own use. Doesn't he have exclusive rights to those modifications so long as he never distributes the modified program? The article establishes that using the code to generate his site does not distribute the code.
It might matter whether Sakic wrote Connolly's code as an employee (I can't remember the legal phrase - "work under contract"?) or as a separate entity who then distributed the code to Connolly.
The GPL says that if I modify a GPLed program *and distribute it*, anyone I distribute it to has the right to the modified source, and to make additional modifications and distributions. This doesn't look to be the situation here.
If we regard Sakic as the creator and copyright holder, then he distribited the code to Connolly. By default, he has the right to also distribute the code elsewhere. Could he be contractually prevented from doing so? Was he contractually prevented from doing so in this case? (If no contract was signed, this might not put him in the clear. Who knows what the 'default' situation is.)
On the other hand, if Connolly's company is the creator and copyright holder (rather than Sakic), then Sakic has broken copyright if he distributed the company's code.
(Reminder: Connolly would have to prove that his code could not legally be redistributed *and* that the code in Mambo contains/is derived from his code. I am only addressing the first issue.)
From the fine article: "As you reduce the size of such a battery, the amount of stored energy goes down exponentially. Reduce each side of a cubic battery by a factor of 10 and you reduce the volume--and therefore the energy you can store--by a factor of 1000."
No, the amount of stored energy goes down polynomially (specifically, cubically), dammit! Must even science articles abuse the word "exponentially"?
"Download the specification, which describes a nonproprietary file format for storing camera raw files that can be used by a wide range of hardware and software vendors." from Adobe'sweb page about DNG. (emphasis added.)
Of course they'd make the level designation open-ended. There's nothing wrong with having level 22's in 10 years.
Otherwise you have to come up with some really nifty technology so that the packaging on a game that says 'needs level 7 computer' will change after 2 years of sitting on a shelf to say 'needs level 5 computer'.
If I were in Microsoft and trying to implement this system, I'd make a utility in the OS that determined the level of your computer. You'd run it, and it would say "This is a level 5 computer". Drill down for details, and you find out you have level 5 RAM, level 6 CPU and level 7 GPU, so you know you can upgrade to a level 6 just by adding RAM.
This would level the playing field between Dell and Compaq: if it says it is a level 5 computer, it is *Microsofts* idea of a level 5.
It would, of course, create a pressure to make hardware that barely qualifies for a particular quality level, and to be optimised for whatever benchmarks MS use for determining level.
"A company doesn't acquire trademark rights simply by choosing a trademark or even by stating its intent do use it. Trademarks... are established by actual use in conjuction with specific goods."
So if I find out that Microsoft are just about to launch a new soup called 'Wibble', I can start selling my own 'Wibble' soup first, then claim they are infringing my trademark when they launch? There must be something that prevents this.
Once (Chicago O'Hare, c1980.) Due to faulty maintenance procedures (now discontinued), lack of locking on slats (now fixed) and engine-out-on-takeoff procedures that sacrificed air speed for altitude.
There are three DC-10 crashes (that I can think of off hand) that could reasonably be blamed at least partially on the design of the plane: we've mentioned two (Paris, Chicago). The third is Sioux City, where an uncontained engine failure in cruise disabled all three hydrolic systems. The plane crash landed with (from memory) about 110 deaths and 180 survivors.
Other planes of similar size and age (Lockheed L1011 tristar, 747) had four hydrolic systems. Had the DC-10 had four *and* (that is a big 'and') the fourth had not been disabled, it is unlikely there would have been any deaths. (A 747 once had 3 out of 4 hydrolic systems disabled on takeoff, and landed safely.)
In terms of safety, I'd be more worried about any model of airplane less than a few years old than I'd be about a well maintained DC-10. Let other people find the surprises first.
There is a rather more extreme case of this with the FAA - when first deployed, the cargo doors of the DC-10 were unsafe, with a failure mode that was likely to make the plane uncontrolable in flight.
This occured in flight, and through luck (which allowed some degree of control) and extraordinary airmanship, the plane was landed safely. (This is known as "The Windsor Incident.")
McDonnell-Douglas didn't want to do a proper redesign of the door mechanism, and the FAA head was a 'companies know best' political appointee, so the result was McD added little windows to the door so that the guy closing the door could look to see it had all engaged properly. (This was over vigourous opposition by the NTSB, who recognized the inadequacy of the fix.)
The situation: A single failure (not looking, or looking but not noticing an unsafe condition) by a non-safety trained close to minimum wage employee could cause the deaths of hundreds of people.
Result: over 300 dead when a Turkish Airlines DC-10 crashed near Paris. The guy who closed the door hadn't even been told he was supposed to check the little windows.
Safety critical systems must be tolerant of human error. If a single omission by a human leads to a hazardous situation, this is primarily the fault of the system, not the human.
Slashdot Mirror
I was involved in a similar, but very much smaller scale, experiment for my MSc thesis (JANZOS), attempting to find detect gamma rays from the (then very recent) supernova 1987A in the Large Magellanic Cloud.
So supernovae were a prime suspect source back then.
We had three (not four) 2 metre (not 12 metre) telescopes with about 30 'pixels' each (compared to a few thousand for HESS.) (I actually worked on another part of the experiment, which used particle detectors to detect higher energy showers.)
A significant problem is to distinguish between showers created by gamma rays and ones created by charged particles (mostly protons.) The charged particle showers are 'uninteresting', because the direction they come from is uncorrelated to their source - they move on curved paths due to galactic magnetic fields. Unfortunately, they are about 99% of the cosmic rays. We were not able to distinguish, so we had a large 'signal to noise' problem.
There was a single telescope similar to these ones in the mid 80s (the Whipple Telescope, I think) which claimed to be able to distinguish by details of shower structure. (We didn't have the resolution, nor perhaps the light gathering power, to make use of this.) I presume HESS has built on this work.
Note that this result does not necessarily tell us about the very highest energy cosmic rays. There is a change in the slope of the spectrum at (from memory) about 10^15 electron volts, so it is likely that different processess are involved on either side of this boundary. I think there were also theoretical reasons to think that supernovae could not accelerate particles to such high energies.
As I recall, the models for acceleration generally required shock waves in a gas with magnetic fields. Particles could repeatedly bounce across the shock, getting accelerated each time. (Think of a ball bouncing between two walls that are moving towards each other.)
Everyone I talk to seems to be "I HATE Kerry", or "I HATE Bush", but no one actually *likes* anyone anymore it seems...
Vote Cthulu 2004: When you're tired of voting for the lesser of two evils.
More seriously, the US presidential election is pretty much the ideal case for a single-transferable-vote electoral system. Just one winner, large incentive for people to study the candidates long enough to make an informed choice.
The horizon limits you to viewing half the sky. Atmospheric effects make it difficult/unviable to view close to the horizon, so in practice this is even more limited - say 1/3 of the sky. In addition, daylight restricts your observing time by a factor of more than two, and for faint diffuse objects (glaxies, nebulae) you also can't observe when a bright moon is in the sky, nor, of course, when it is cloudy - so maybe you end up with 4 hours per day of good observing time per night, on average. A space telescope suffers none of these limitations. (Well, just a little - you can't observe too close to the sun, moon or earth.)
However, although you can only observe 1/3 of the sky at a given moment, the motion of the stars through the night and year means you can observe much more of the sky if you're prepared to wait. Furthermore, if the telescope costs a small fraction of the cost of a space telescope, you can build many of them in different parts of the world, to overcome these limitations.
There are other reasons for going into space - atmospheric bluring, absorption and emission.
No - there is lots that Hubble can do that this telescope can't, and lots that no earthbound telescope can do. (See my other comment on this article: #10553382
It is another question of whether that extra capability is worth the extra cost.
The story gives the impression that the LBT will completely replace Hubble, and do a better job, while being vastly cheaper.
This is an overstatement. There is lots that Hubble can do that no other telescope can, being a unique combination of aperature (light gathering power and resolution), instruments (many wavelengths, imaging and spectroscopic) and being above the atmosphere (no 'seeing', no atmospheric absorption or emission in UV and IR.)
(This is not to downplay the LBT - doing better than HST in some aspects, and as well but much cheaper in others, is very valuable.)
Having quickly scanned the website for this telescope, I can't see how they are counteracting the bluring of 'seeing' (atmospheric turbulance). It is inconceivable that they have neglected it, but I don't see where. Adaptive optics can help, but have limitations of their own.
Another limitation of the LBT is that the high resolution reconstruction will require 3 observations at different times - so it only works well with non-time-varying targets. This is a minor limitation, however - a large majority of targets for which you want high resolution are non-variable.
(IWAA: I was an astonomer. PhD, but no further.)
Astronomers have occasionally used balloon borne telescopes for getting above most of the atmosphere, as it is much cheaper than a satellite. If there is a mass-produced long duration stratespheric balloon/airship available, it could make this much more viable.
As an aside - the article also discusses "Terahertz imaging." One terahertz corresponds to wavelength of about 0.3 mm or 300 microns - extreme IR, or short sub-millimetre, depending on your point of view.
This isn't yet-another-varient of SQL - it is a from-the-ground-up replacement. Think Perl vs Python, not K&R C vs Borland C.
(Disclosure: I've programmed SQL, Perl and K&R C, and read Darwin and Date, but I have not programmed Python, Borland C or Tutorial D.)
Pedantic quibble time: Actually it is >1400 times as powerful as fusion, and even more compared to fission.
The nuclear reaction that releases the greatest percentage of mass into energy is 4 protons (hydrogen nuclei) -> 1 alpha particle (helium nucleus.) This has an 'efficiency' of about 0.7%.
This is also the reaction that powers the sun. (Technically, it is a series of reactions - either the 'p-p chain' or the 'CNO cycle', but that is too much detail to go into here.)
In principle, the best you could do would be 56 protons -> iron nucleus, but this wouldn't be feasible.
To be fair, so far as I know we haven't every built a facility dedicated to efficient large-scale production and collection of anti-matter. It wouldn't surprise me if such a facility reduced the cost by a factor of 100 or 1000. I expect that a rough design for such a facility is a goal of this research.
If we assume that the maximum cost the military would pay for enough antimater to blow up a building was $1 million, the price has to come down by a factor of 3,000,000. I really doubt this is feasible in the forseeable future (say 50 years.)
No, no, it really is a Turin machine. This is a device with internal states and a shroud that it moves back and forth, writing and removing mysterious negative images of dead people.
Don't panic/celebrate in anticipation of antimatter weapons being deployed 15 years from now.
From the article:
"about 50-millionths of a gram could generate a blast equal to the explosion (roughly 4,000 pounds of TNT, according to the FBI) at the Alfred P. Murrah Federal Building in Oklahoma City in 1995."
and
"With present techniques, the price tag for 100-billionths of a gram of antimatter would be $6 billion"
from which we can calculate that blowing up a building with antimatter will cost about 3 trillion dollars. (And tens or hundreds of millions for the equipment to confine the antimatter until you want it to detonate, but that is negligible in comparison.)
Also notice that while the antimatter may be the ultimately compact explosive, the containment equipment required will increase the size of a warhead by many orders of magnitude. No antimatter rifle bullets anytime soon.
These days, it's not the Iraqi government that's kidnapping, torturing, and murdering people, but a group of loosely-affiliated amateurs.
Come on, the US army isn't quite that poorly organized.
(Sorry, I couldn't resist.)
Assuming all this is so, I think a final step is required in the chain. The existence of people who can legally distribute Connolly's code does not imply that Sakic has a legally distributable copy.
Sakic needs to get a copy of his code from (directly or indirectly) one of the 'registered users', who, via the GPL, have the right to redistribute it.
This may require some persuading and legal indemnification, as the 'registered users' know that Connolly is litigation-happy and may come after them.
Possibly Sakic would need to resubmit his controversial Mambo contribution after receiving this code - ask a lawyer.
Off course, all this is moot if the Mambo code is not derivative of the Connolly code.
OK, here is a hypothetical situation:
Alice is a contractor. She modifies a GPL program under contract for BigCorp. She sends a copy of the modifications to BigCorp.
Alice has distributed the code to BigCorp, so she has no legal right to complain should BigCorp distribute it further, so long as BigCorp do so under the GPL.
Now say Alice's contract says that she will not distribute her modifications to anyone else. Is this legal/enforcable? Did Alice have the right to modify the code while bound by such a restriction?
I independently came to (and posted, although less succinctly) the same conclusions as jhoger.
Anonymous ad-hominem attacks vs logical argument - hm, which should I belive?
I suppose I've just become an astroturfer as well.
The article makes two claims: one is that the two pieces of code are different. I'm not disputing this.
The other is that even if they were the same, the GPL makes it OK. I am disputing this.
I.e. the article puts forth two planks: if either hold up, Mambo is safe. I think one of those planks is rotten.
(IANAL)
"Copyright infringement is a broader term and would seem to be a better fit for this situation; after all, the contract called for all copyrights to be assigned to Furthermore, Inc. But the code was a derivative of GPL-licensed code, thereby making it a derivative work of the parent code and automatically licensing it under the GNU General Public License."
But doesn't the GPL only take effect if the code is distributed? Connolly employed someone to modify GPLed code, for his own use. Doesn't he have exclusive rights to those modifications so long as he never distributes the modified program? The article establishes that using the code to generate his site does not distribute the code.
It might matter whether Sakic wrote Connolly's code as an employee (I can't remember the legal phrase - "work under contract"?) or as a separate entity who then distributed the code to Connolly.
The GPL says that if I modify a GPLed program *and distribute it*, anyone I distribute it to has the right to the modified source, and to make additional modifications and distributions. This doesn't look to be the situation here.
If we regard Sakic as the creator and copyright holder, then he distribited the code to Connolly. By default, he has the right to also distribute the code elsewhere. Could he be contractually prevented from doing so? Was he contractually prevented from doing so in this case? (If no contract was signed, this might not put him in the clear. Who knows what the 'default' situation is.)
On the other hand, if Connolly's company is the creator and copyright holder (rather than Sakic), then Sakic has broken copyright if he distributed the company's code.
(Reminder: Connolly would have to prove that his code could not legally be redistributed *and* that the code in Mambo contains/is derived from his code. I am only addressing the first issue.)
Is there a copyright lawyer in the house?
No, it is not valid. "Exponentially" means something like 3^x, not like x^3.
Claiming they are the same is like claiming cows are carnivores, because they one of the participants in an animal-eats-animal relationship.
From the fine article: "As you reduce the size of such a battery, the amount of stored energy goes down exponentially. Reduce each side of a cubic battery by a factor of 10 and you reduce the volume--and therefore the energy you can store--by a factor of 1000."
No, the amount of stored energy goes down polynomially (specifically, cubically), dammit! Must even science articles abuse the word "exponentially"?
"Download the specification, which describes a nonproprietary file format for storing camera raw files that can be used by a wide range of hardware and software vendors." from Adobe'sweb page about DNG. (emphasis added.)
Of course they'd make the level designation open-ended. There's nothing wrong with having level 22's in 10 years.
Otherwise you have to come up with some really nifty technology so that the packaging on a game that says 'needs level 7 computer' will change after 2 years of sitting on a shelf to say 'needs level 5 computer'.
If I were in Microsoft and trying to implement this system, I'd make a utility in the OS that determined the level of your computer. You'd run it, and it would say "This is a level 5 computer". Drill down for details, and you find out you have level 5 RAM, level 6 CPU and level 7 GPU, so you know you can upgrade to a level 6 just by adding RAM.
This would level the playing field between Dell and Compaq: if it says it is a level 5 computer, it is *Microsofts* idea of a level 5.
It would, of course, create a pressure to make hardware that barely qualifies for a particular quality level, and to be optimised for whatever benchmarks MS use for determining level.
There is an interesting bit in there:
... are established by actual use in conjuction with specific goods."
"A company doesn't acquire trademark rights simply by choosing a trademark or even by stating its intent do use it. Trademarks
So if I find out that Microsoft are just about to launch a new soup called 'Wibble', I can start selling my own 'Wibble' soup first, then claim they are infringing my trademark when they launch? There must be something that prevents this.
Once (Chicago O'Hare, c1980.) Due to faulty maintenance procedures (now discontinued), lack of locking on slats (now fixed) and engine-out-on-takeoff procedures that sacrificed air speed for altitude.
There are three DC-10 crashes (that I can think of off hand) that could reasonably be blamed at least partially on the design of the plane: we've mentioned two (Paris, Chicago). The third is Sioux City, where an uncontained engine failure in cruise disabled all three hydrolic systems. The plane crash landed with (from memory) about 110 deaths and 180 survivors.
Other planes of similar size and age (Lockheed L1011 tristar, 747) had four hydrolic systems. Had the DC-10 had four *and* (that is a big 'and') the fourth had not been disabled, it is unlikely there would have been any deaths. (A 747 once had 3 out of 4 hydrolic systems disabled on takeoff, and landed safely.)
In terms of safety, I'd be more worried about any model of airplane less than a few years old than I'd be about a well maintained DC-10. Let other people find the surprises first.
There is a rather more extreme case of this with the FAA - when first deployed, the cargo doors of the DC-10 were unsafe, with a failure mode that was likely to make the plane uncontrolable in flight.
This occured in flight, and through luck (which allowed some degree of control) and extraordinary airmanship, the plane was landed safely. (This is known as "The Windsor Incident.")
McDonnell-Douglas didn't want to do a proper redesign of the door mechanism, and the FAA head was a 'companies know best' political appointee, so the result was McD added little windows to the door so that the guy closing the door could look to see it had all engaged properly. (This was over vigourous opposition by the NTSB, who recognized the inadequacy of the fix.)
The situation: A single failure (not looking, or looking but not noticing an unsafe condition) by a non-safety trained close to minimum wage employee could cause the deaths of hundreds of people.
Result: over 300 dead when a Turkish Airlines DC-10 crashed near Paris. The guy who closed the door hadn't even been told he was supposed to check the little windows.
Safety critical systems must be tolerant of human error. If a single omission by a human leads to a hazardous situation, this is primarily the fault of the system, not the human.