There are some things where its not clear whether they are a physical object or not. Clearly they are talking about the master pattern for the dovetails. A dovetailer is a machine which cuts the end of a bit of wood into crennelations (up and down bits like on the top of a castle). It does this often by using a hand router to cut around a metal form which is the "master form" which then produces copies of the master form in wood. Obviously they think that their master form is an original pattern, and that copying is is analogous to photocopying a book. That is: they consider the pattern, rather than the metal of the form to be the thing of value. You could equally use the pattern of the metal to cut copies of that pattern into metal, rather than wood; thus destroying what they consider to be the value of the item: its rarity and uniqueness.
Having seen a few dovetail forms, I'd have to say it would want to be a pretty amazing form.
I think another way of looking at it would be to consider keys. your landlord can't stop you from lending your key to someone else, but I'd think that making copies of it and distributing it to all of your friends would create a bit of a stir, as would selling your key: and the key is also one of your "rivalrous assets".
Except that you wasted two days essentially redoing what was already a finished project. The staples of FORTRAN are longitivity, usability and speed. Think about this: pretty much every FORTRAN code ever written can be compiled with a current complient f90 or f95 compiler. And this includes the pre fortran 66 stuff which is still amazingly abundant in the scientific world.
If you want FORTRAN without the bad habits use Pascal.
no, it has to be: a) A small part of a code that comprises a significant part of the company's business b) Code which has been integrated into a GPL product. c) The company also sells a Linux distribution. d) The code is part of the kernel, which the company can't touch, or risk becoming uncompliant with standards=no longer Linux=no longer saleable.
Which is exactly what SCO claims is their problem. Look, I don't agree with them. But if I was CEO of a company considering supporting Linux, and your hardline option was really the law, I'd not support Linux. I therefore think that, while you have a point, that a softer line to the all-or-nothing GPL argument is far more likely to benefit Linux in the long run. Largely because at least some of the options we're seeing for making a profit with open source involve mixing GPL and non-GPL software. That means that we as a community need to reduce the number of irreversible mistakes a company can make when dealing with a mix of GPL and proprietry solutions.
So you're saying, if I steal your code, and put it in the Linux kernel, so that it is a vital part of the kernel. Then you form a company selling both that code, and a linux distribution. that when I tell you about the code I stole, that you immediately have no income, through my illegal action because:
a) You can't sell your original code because you don't want to open the modifications to what is now a GPLed code. b) You can't sell the distribution any more because it's suddenly a mix of GPL and non-GPL code.
Great! Suddenly any firm can be destroyed by stealing a small section of code and GPLing it.
Now I know you're starting from the assumption that SCO is wrong, but handled wrongly, this case could destroy commercial Linux as we know it.
The current Linux kernel has what, 2.5 million lines of code? Assuming a coder can check one line per second it will take about 700 hours to check. Or about 30,000 USD. Now, do you think that it is reasonable to have a "Linux tax" of 30000 USD on any conpany which wants to have anything to do with the Kernel?
What if this number is more like 300,000? 3 million? At some point, it is not reasonable to check each line of code. The question is on which side of that point SCO was when they started to sell Linux.
Yeah, the analogy sucked. But the point was that simple reselling doesn't constitute approval. I think that this is enormously important because it potentially highlights a huge drawback for businesses who want to use GPLed software. If they can't take back the accidental GPLing of something when they discover it, then it greatly increases the danger of the GPL to business, regardless of its other advantages.
What if they did stop selling the software? Then they have no income while the legal mills slowly grind, leaving a non-legal win option for IBM. IANAL, but I think that you'd find a good argument that the decision as to whether to sell the software or not should be made at the close, not at the beginning of the suit, since it can be that they lose the suit, in which case they didn't have to stop selling the software. A very nice catch-22, which we'd be wining about if it was RedHat vs Microsoft.
The code was open for all to see: in the cellar with no lights or stairs in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying "Beware of The Leopard".'"
How many million lines of code does an ordinary linux distribution have now? Claiming it was open is like expecting each and every customer to read every page of a 70 page EULA. Nice in theory. Not practible.
I don't understand this logic. Lets say I work for a company making widget subcomponents. I also retail widgets made by a range of companies. One of those companies stole one of my widget subcomponents and without my knowledge incorporated it into their widget, which was subsequently retailed by my company.
How exactly is the retail of the dodgy widget an excuse for the thief?
You know, I've never really understood this sort of comment on slashdot. I mean, we're geeks, right. I mean, something like 90% of the stories on this site revolve around how perfection in automation is practically impossible to achieve. About how, a certain level of bugs is to be expected no matter how much the software is tested.
What, exactly, makes you think that NASA has some sort of secret magic bullet that they're not telling us about? What's the reason for the space shuttles carrying astronauts? Because automation is unreliable, at best.
The scientists at NASA don't just sorta guess. They make educated guesses. Sometimes those guesses are wrong. The stuff we don't know about flight at the kind of speeds would shock you. (Try googling for "real gas effects" or "radiation heat transfer" together with "re-entry" if you're interested.) For instance I believe that on the first shuttle flight the prediction of center of lift was off by 0.7%, necessitating doubling the flap area.
So combine the science we don't fully understand with automation and we will have failures. It's just a fact. Would you prefer they didn't try?
The literacy factor: Adults are not reading as often as they used to, and the literacy level of adult fiction has lowered over the decades to keep up with the times. Compare a newspaper from a hundred years ago to one today. Compare a popular novel to one today. People like easier material. They get lazy. So when they pick up a HP book they find it's easy to read, contains a good story, and they feel as if they are accomplishing something.
I can't agree with this. I think it's one of the major problems with literate society that what is pushed as "good" is not what most people want to read. I read a lot of fiction. It's a hobby I've certainly gone through stages where I was reading more than 1000 books per year. I have read a good cross section of classics too, but they're generally the ones I don't read because they're too hard to be fun to read.
People who read a lot can get through Dostoyevsky or Joyce, but someone who doesn't read a lot can't. I would go so far as to say that in a lot of books, there are streaks of brilliance surrounded by large wastelands of mediocrity. Some people can wade through the filler easily, but if you're a slow reader, a week of mediocrity is too much. However what is preserved is the books which experienced readers want to read. Writing 100 years ago was much like writing today pick up a copy of the essays of Ben Franklin if you'd like an example of writing pitched to a broader audience. Likewise newspapers- Pick up a copy of The Times or even The Guardian, and I think you'll find that your comments about literary standards have as much to do with US publishers pitching to the lowest common denominator as about mean levels of literacy .
Think about this: If you have to recommend to someone who reads one book a year something that they'll enjoy, what would you recommend? I'd say you could do worse than Harry Potter.
Think about it. In English, there are two completely different sets of grammar for dealing with singular "one" and plural "two". We use the same grammer for differentiating between varying degrees of plurality. Thus it seems likely that the concept of singular and plural is a fundamental lingustic idea.
The words small/big and live/die are antonyms, but think about how you'd use them in sentences, and consider that the list is meant to be multilingual. "where are you living?" "where are you dying?" "how small is your house?" "how big is your house?"
English has a different grammer to refer to time displacement and quantity than position and number. "how long did it take?" "how far was it?" "how many were there?"
Re:1 accident in 62.5 flights IS acceptable
on
Shuttle Politics
·
· Score: 1
Clearly the solution is to upgrade all the astro-naughts to astro-ones.
This upgrade should licenced under the GPL, and free (as in beer). Come to think of it I think I'll follow that upgrade path too. Off to the pub!
I've wondered a lot about the accuracy problem. Ever seen a mirage? It's light being bent from the sky through a density gradient near the ground.
So what happens if I fire my laser gun at a mirage? Well, some of the beam goes to the target, and some to the sky. I'd guess, with this example that laser weapons are not the general solution to accuracy that you're thinking.
What about some other conditions where laser weapons are likly to suffer: Rain, Over water, windows in urban conflict sandstorms, jungles, fog.
I see a lot of disadvantages to this weapon. And not a lot of advantages. Lasers are generally poor converters of energy to light. So that means carrying a lot of energy. They have poisonous substances inside. They have rare substances inside, all of which lead to supply problems.
In addition, These will not be silent, and it begs the question of how being recoilless is going to repay the significant research commitment involved.
Look, That's what wind tunnels are _for_. If something you do in a wind tunnel is not the same as in free flight, then the wind tunnel tests are worthless.
A typical time constant for hydrogen scramjet engines is 100 microseconds. That is to say, that if your test time is longer than that, say a few milliseconds, then you have long-duration flight.
You're right to say that the stability of the combustion system is a problem in scramjets, but it's not as big a discovery as you'd think. Since the combustion system is the scramjet (There is effectively nothing else in a scramjet) then saying that there's a problem with the stability of the combustion system is not isolating your problem.
Ignition is no problem. Fuel mixing is a problem. Intakes are a problem. Boundary layer separation is a problem. Stable operation at high equivalence ratios is a problem. Heat transfer in the combustion chanber is a problem.
Are all the real rocket scientists just incompetant morons. Are the designers of the Space shuttle, CRV and Soyuz just idiots that they're attaching all this useless flight control and power stuff on re-entry?
If a scramjet has the shock pushed out the front, most designs simply spill the shock so that the pressure in the combustion chamber is the stagnation pressure of the flow, plus the additional pressure due to the equilibrium reaction of the fuel. You can design for this.
Scramjets have been designed which will take pitch/yaw of +-4 degrees. That doesn't sound like much, but remember that you're going pretty danm fast.
Nuclear rockets work by superheating steam. It's not radioactive. The problem is when a nuclear rocket explodes.
"Nobody's figured out how to keep a scramjet lit.."
Think about this: That Australian free flight experiment was building on 15 years on research in Australia, and they launched it because they were 100% sure that they knew how to keep a scramjet lit in a ground test.
There are atr least 4 commom methods which you can find in the literature. Cavity, step, shock induced combustion, high temp homogenous flow. and that's just off the top of my head.
You can forget about getting a Nobel prize for a better cigarette lighter.
Explain to me again why a robot would need to build an oxygen scrubber?
One of the factors which argues for using robots is simplicity. Not of the robot itself, but all of that stuff which breaks, and is heavy: Pressure shell Oxygen system water system room to move exercise equipment Is not needed for a robot.
Let me do a thought experiment. Take those 7 astronauts. Say they have a (conservative) combined weight of 560kg. Say in addition that all the life support stuff weighs about the same.
Those astronauts (7 people) plus trainers (say, 20 people) have trained for (say) 1 year to get on that mission.
You can substitute your own numbers.
Take then those 27 people, and give them 1 year, and 1 tonne to build a robot. Could it do 90% of what those astronauts could do?
There's only one problem with your argument. We don't know all about ablative shields. Basically, the programs you mention had a factor 10 uncertainty in the recession rate of the ablative system. That's fine, but it means that for every 1kg of ablative you need, you have to carry 9kg with you to be safe.
That's inefficient.
Research into ablatives is continuing, but it would be a bad idea to assume that the Engineers at NASA are incompetent.
Wave drag is a problem with hypersonic vehicles, which is why you tailor your intake and exhaust to each other. See the Busemann biplane for the classic example (Google is your friend!).
Viscous drag make up about half the drag of current designs, and most of this is in the combustion chamber. This is because the air is compressed in the combustion chamber, and drag is proportional to density. Therefore, viscous drag and peak heating loads can be reduced by reducing the compression ratio in the intake.
A more accurate example of the idea of a scramjet using your numbers would be:
Rocket A thrusts at 100lbs, weighs 10 lbs, and has a payload of 2 oz.
Scramjet B thrusts at 100lbs, weighs 1 lb, and has a payload of 2oz. The thrust is only 1/5, but the total mass is 1/10, so it accelerates the payload twice as fast.
Unfortunately, nobody has demonstrated any net thrust in a scramjet in net flight, or in flight conditions in a wind tunnel (as far as I'm aware), so the net thrust of scramjet B is actually zero, which makes the whole point moot at the present time.
An interesting thesis. How many people are affected if there are 10% too few teachers trained this year? How many if their company goes belly-up taking their pension plans with it? How many if the guy who installs the brakes on your car does it wrong?
What I'm getting at is that responsibility, and consequences for responsibility are pretty relative things. Very few engineers actually find themselves in the situation of designing anything. Most who do design, use "safety factors", also known as fuck-up factors of between 4 and 10. That is, so long as your calculation is not off by more than the safety factor, nobody will ever know. A very few engineers design with safety factors under 2. These are called Rocket Scientists.
I don't particularly see how designing something with a safety factor of 10 is more responsibility than being in charge of a pension fund for 10000 people.
The efficiency of your gun, for a given pressure is a function of the speed of sound of your driver gas. Better guns are made with light gases (He, H2) or hot gases (combustion products).
If your CO2 was working better than air, it was because you had a higher pressure.
Slashdot Mirror
There are some things where its not clear whether they are a physical object or not. Clearly they are talking about the master pattern for the dovetails. A dovetailer is a machine which cuts the end of a bit of wood into crennelations (up and down bits like on the top of a castle). It does this often by using a hand router to cut around a metal form which is the "master form" which then produces copies of the master form in wood. Obviously they think that their master form is an original pattern, and that copying is is analogous to photocopying a book. That is: they consider the pattern, rather than the metal of the form to be the thing of value. You could equally use the pattern of the metal to cut copies of that pattern into metal, rather than wood; thus destroying what they consider to be the value of the item: its rarity and uniqueness.
Having seen a few dovetail forms, I'd have to say it would want to be a pretty amazing form.
I think another way of looking at it would be to consider keys. your landlord can't stop you from lending your key to someone else, but I'd think that making copies of it and distributing it to all of your friends would create a bit of a stir, as would selling your key: and the key is also one of your "rivalrous assets".
Except that you wasted two days essentially redoing what was already a finished project. The staples of FORTRAN are longitivity, usability and speed. Think about this: pretty much every FORTRAN code ever written can be compiled with a current complient f90 or f95 compiler. And this includes the pre fortran 66 stuff which is still amazingly abundant in the scientific world.
If you want FORTRAN without the bad habits use Pascal.
It's security through obscurity....
no, it has to be:
a) A small part of a code that comprises a significant part of the company's business
b) Code which has been integrated into a GPL product.
c) The company also sells a Linux distribution.
d) The code is part of the kernel, which the company can't touch, or risk becoming uncompliant with standards=no longer Linux=no longer saleable.
Which is exactly what SCO claims is their problem. Look, I don't agree with them. But if I was CEO of a company considering supporting Linux, and your hardline option was really the law, I'd not support Linux. I therefore think that, while you have a point, that a softer line to the all-or-nothing GPL argument is far more likely to benefit Linux in the long run. Largely because at least some of the options we're seeing for making a profit with open source involve mixing GPL and non-GPL software. That means that we as a community need to reduce the number of irreversible mistakes a company can make when dealing with a mix of GPL and proprietry solutions.
Remember: Embrace and Extend!
So you're saying, if I steal your code, and put it in the Linux kernel, so that it is a vital part of the kernel. Then you form a company selling both that code, and a linux distribution. that when I tell you about the code I stole, that you immediately have no income, through my illegal action because:
a) You can't sell your original code because you don't want to open the modifications to what is now a GPLed code.
b) You can't sell the distribution any more because it's suddenly a mix of GPL and non-GPL code.
Great! Suddenly any firm can be destroyed by stealing a small section of code and GPLing it.
Now I know you're starting from the assumption that SCO is wrong, but handled wrongly, this case could destroy commercial Linux as we know it.
The current Linux kernel has what, 2.5 million lines of code? Assuming a coder can check one line per second it will take about 700 hours to check. Or about 30,000 USD. Now, do you think that it is reasonable to have a "Linux tax" of 30000 USD on any conpany which wants to have anything to do with the Kernel?
What if this number is more like 300,000? 3 million? At some point, it is not reasonable to check each line of code. The question is on which side of that point SCO was when they started to sell Linux.
So you're claiming that by not reading a 70 page EULA and agreeing in good faith, that you can agree to sell your first born child?
I think that this is a another point hanging on the legal definition of "reasonable".
Yeah, the analogy sucked. But the point was that simple reselling doesn't constitute approval. I think that this is enormously important because it potentially highlights a huge drawback for businesses who want to use GPLed software. If they can't take back the accidental GPLing of something when they discover it, then it greatly increases the danger of the GPL to business, regardless of its other advantages.
What if they did stop selling the software? Then they have no income while the legal mills slowly grind, leaving a non-legal win option for IBM. IANAL, but I think that you'd find a good argument that the decision as to whether to sell the software or not should be made at the close, not at the beginning of the suit, since it can be that they lose the suit, in which case they didn't have to stop selling the software. A very nice catch-22, which we'd be wining about if it was RedHat vs Microsoft.
The code was open for all to see:
in the cellar
with no lights
or stairs
in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying "Beware of The Leopard".'"
How many million lines of code does an ordinary linux distribution have now? Claiming it was open is like expecting each and every customer to read every page of a 70 page EULA. Nice in theory. Not practible.
I don't understand this logic. Lets say I work for a company making widget subcomponents. I also retail widgets made by a range of companies. One of those companies stole one of my widget subcomponents and without my knowledge incorporated it into their widget, which was subsequently retailed by my company.
How exactly is the retail of the dodgy widget an excuse for the thief?
You know, I've never really understood this sort of comment on slashdot. I mean, we're geeks, right. I mean, something like 90% of the stories on this site revolve around how perfection in automation is practically impossible to achieve. About how, a certain level of bugs is to be expected no matter how much the software is tested.
What, exactly, makes you think that NASA has some sort of secret magic bullet that they're not telling us about? What's the reason for the space shuttles carrying astronauts? Because automation is unreliable, at best.
The scientists at NASA don't just sorta guess. They make educated guesses. Sometimes those guesses are wrong. The stuff we don't know about flight at the kind of speeds would shock you. (Try googling for "real gas effects" or "radiation heat transfer" together with "re-entry" if you're interested.) For instance I believe that on the first shuttle flight the prediction of center of lift was off by 0.7%, necessitating doubling the flap area.
So combine the science we don't fully understand with automation and we will have failures. It's just a fact. Would you prefer they didn't try?
Once the air runs out, the humans also become multi-million dollar paperweights!
The literacy factor: Adults are not reading as often as they used to, and the literacy level of adult fiction has lowered over the decades to keep up with the times. Compare a newspaper from a hundred years ago to one today. Compare a popular novel to one today. People like easier material. They get lazy. So when they pick up a HP book they find it's easy to read, contains a good story, and they feel as if they are accomplishing something. I can't agree with this. I think it's one of the major problems with literate society that what is pushed as "good" is not what most people want to read. I read a lot of fiction. It's a hobby I've certainly gone through stages where I was reading more than 1000 books per year. I have read a good cross section of classics too, but they're generally the ones I don't read because they're too hard to be fun to read. People who read a lot can get through Dostoyevsky or Joyce, but someone who doesn't read a lot can't. I would go so far as to say that in a lot of books, there are streaks of brilliance surrounded by large wastelands of mediocrity. Some people can wade through the filler easily, but if you're a slow reader, a week of mediocrity is too much. However what is preserved is the books which experienced readers want to read. Writing 100 years ago was much like writing today pick up a copy of the essays of Ben Franklin if you'd like an example of writing pitched to a broader audience. Likewise newspapers- Pick up a copy of The Times or even The Guardian, and I think you'll find that your comments about literary standards have as much to do with US publishers pitching to the lowest common denominator as about mean levels of literacy . Think about this: If you have to recommend to someone who reads one book a year something that they'll enjoy, what would you recommend? I'd say you could do worse than Harry Potter.
Think about it. In English, there are two completely different sets of grammar for dealing with singular "one" and plural "two". We use the same grammer for differentiating between varying degrees of plurality. Thus it seems likely that the concept of singular and plural is a fundamental lingustic idea.
The words small/big and live/die are antonyms, but think about how you'd use them in sentences, and consider that the list is meant to be multilingual.
"where are you living?"
"where are you dying?"
"how small is your house?"
"how big is your house?"
English has a different grammer to refer to time displacement and quantity than position and number.
"how long did it take?"
"how far was it?"
"how many were there?"
Clearly the solution is to upgrade all the astro-naughts to astro-ones.
This upgrade should licenced under the GPL, and free (as in beer). Come to think of it I think I'll follow that upgrade path too. Off to the pub!
I've wondered a lot about the accuracy problem. Ever seen a mirage? It's light being bent from the sky through a density gradient near the ground.
So what happens if I fire my laser gun at a mirage? Well, some of the beam goes to the target, and some to the sky. I'd guess, with this example that laser weapons are not the general solution to accuracy that you're thinking.
What about some other conditions where laser weapons are likly to suffer:
Rain, Over water, windows in urban conflict
sandstorms, jungles, fog.
I see a lot of disadvantages to this weapon. And not a lot of advantages. Lasers are generally poor converters of energy to light. So that means carrying a lot of energy. They have poisonous substances inside. They have rare substances inside, all of which lead to supply problems.
In addition, These will not be silent, and it begs the question of how being recoilless is going to repay the significant research commitment involved.
Look, That's what wind tunnels are _for_. If something you do in a wind tunnel is not the same as in free flight, then the wind tunnel tests are worthless.
A typical time constant for hydrogen scramjet engines is 100 microseconds. That is to say, that if your test time is longer than that, say a few milliseconds, then you have long-duration flight.
You're right to say that the stability of the combustion system is a problem in scramjets, but it's not as big a discovery as you'd think. Since the combustion system is the scramjet (There is effectively nothing else in a scramjet) then saying that there's a problem with the stability of the combustion system is not isolating your problem.
Ignition is no problem. Fuel mixing is a problem. Intakes are a problem. Boundary layer separation is a problem. Stable operation at high equivalence ratios is a problem. Heat transfer in the combustion chanber is a problem.
See what I'm getting at?
Are all the real rocket scientists just incompetant morons. Are the designers of the Space shuttle, CRV and Soyuz just idiots that they're attaching all this useless flight control and power stuff on re-entry?
If a scramjet has the shock pushed out the front, most designs simply spill the shock so that the pressure in the combustion chamber is the stagnation pressure of the flow, plus the additional pressure due to the equilibrium reaction of the fuel. You can design for this.
Scramjets have been designed which will take pitch/yaw of +-4 degrees. That doesn't sound like much, but remember that you're going pretty danm fast.
Nuclear rockets work by superheating steam. It's not radioactive. The problem is when a nuclear rocket explodes.
"Nobody's figured out how to keep a scramjet lit.."
Think about this: That Australian free flight experiment was building on 15 years on research in Australia, and they launched it because they were 100% sure that they knew how to keep a scramjet lit in a ground test.
There are atr least 4 commom methods which you can find in the literature. Cavity, step, shock induced combustion, high temp homogenous flow. and that's just off the top of my head.
You can forget about getting a Nobel prize for a better cigarette lighter.
Explain to me again why a robot would need to build an oxygen scrubber?
One of the factors which argues for using robots is simplicity. Not of the robot itself, but all of that stuff which breaks, and is heavy:
Pressure shell
Oxygen system
water system
room to move
exercise equipment
Is not needed for a robot.
Let me do a thought experiment. Take those 7 astronauts. Say they have a (conservative) combined weight of 560kg. Say in addition that all the life support stuff weighs about the same.
Those astronauts (7 people) plus trainers (say, 20 people) have trained for (say) 1 year to get on that mission.
You can substitute your own numbers.
Take then those 27 people, and give them 1 year, and 1 tonne to build a robot. Could it do 90% of what those astronauts could do?
I think so.
There's only one problem with your argument. We don't know all about ablative shields. Basically, the programs you mention had a factor 10 uncertainty in the recession rate of the ablative system. That's fine, but it means that for every 1kg of ablative you need, you have to carry 9kg with you to be safe.
That's inefficient.
Research into ablatives is continuing, but it would be a bad idea to assume that the Engineers at NASA are incompetent.
I'm also working in the field.
Wave drag is a problem with hypersonic vehicles, which is why you tailor your intake and exhaust to each other. See the Busemann biplane for the classic example (Google is your friend!).
Viscous drag make up about half the drag of current designs, and most of this is in the combustion chamber. This is because the air is compressed in the combustion chamber, and drag is proportional to density. Therefore, viscous drag and peak heating loads can be reduced by reducing the compression ratio in the intake.
A more accurate example of the idea of a scramjet using your numbers would be:
Rocket A thrusts at 100lbs, weighs 10 lbs, and has a payload of 2 oz.
Scramjet B thrusts at 100lbs, weighs 1 lb, and has a payload of 2oz. The thrust is only 1/5, but the total mass is 1/10, so it accelerates the payload twice as fast.
Unfortunately, nobody has demonstrated any net thrust in a scramjet in net flight, or in flight conditions in a wind tunnel (as far as I'm aware), so the net thrust of scramjet B is actually zero, which makes the whole point moot at the present time.
An interesting thesis.
How many people are affected if there are 10% too few teachers trained this year?
How many if their company goes belly-up taking their pension plans with it?
How many if the guy who installs the brakes on your car does it wrong?
What I'm getting at is that responsibility, and consequences for responsibility are pretty relative things. Very few engineers actually find themselves in the situation of designing anything. Most who do design, use "safety factors", also known as fuck-up factors of between 4 and 10. That is, so long as your calculation is not off by more than the safety factor, nobody will ever know. A very few engineers design with safety factors under 2. These are called Rocket Scientists.
I don't particularly see how designing something with a safety factor of 10 is more responsibility than being in charge of a pension fund for 10000 people.
The efficiency of your gun, for a given pressure is a function of the speed of sound of your driver gas. Better guns are made with light gases (He, H2) or hot gases (combustion products).
If your CO2 was working better than air, it was because you had a higher pressure.