Following several links from that I reach the conclusion that: a) If you aren't developing for a Mac, or other Apple machine, you probably shouldn't bother with Objective-C, and b) You can pass multiple parameters by defining a function which takes multiple named parameters, and using the names of the parameters each time you call. The names of the parameters will be considered a part of the name of the function. c) And if you happen to define a numeric type, like, say, complex numbers, forget about using +, -, *,/, etc. for the arithmetic operations. You need to call it with names like add:, sub:, mul:, div:, etc.
It may have some advantages, but if so they were never displayed in the tutorials. (Actually, there appears to be a dialect of Objective C for the Mac [it wasn't stated that it was available elsewhere] which implements properties. I.e., getter and setter functions that can be addressed by the naked name of the variable. That's an improvement over C++ and Java. Not over many other languages, such as D and Python.)
To me it appears that Objective C/C++ has missed the boat. It appears to be falling behind various languages that are almost standing still. (I'm *not* impressed by it's way of allowing a function to accept multiple arguments. In Smalltalk that almost works, but Smalltalk has a much cleaner syntax than does Objective C. (And even in Smalltalk I find it bothersome.)
And programmers also very frequently know the results of what they're doing. If you want to blame the quants, go ahead. But be prepared to be bitten by the same tooth.
You KNOW that you didn't run enough tests, or that the program failed some of them. And you shipped it anyway. (And if you didn't know, you should have known.)
I've been spared that, because my company usually WAS my end user. Sometimes I released beta, but it was marked that way, and not used for anything crucial. Then it was fixed as requests came in. But there's no way the programmers in commercial houses can't KNOW that they haven't tested things enough. But if they do, they'll be outsold by the competition, because the customers can't tell the difference...until after, sometimes long after, they've bought it.
It didn't have any measurable chance of working, so playing "What if it did work?" is meaningless unless you're writing an alternate history novel. And then it means you'd do better to pick a different altered history. E.g., what if the US had supported the Geneva Conference on Viet Nam? That one has not only some plausibility, but also some potentially quite interesting consequences.
(OTOH, "What if the Germans won WWII?" has been done frequently, to profound artistic reaction, despite the fact that after Germany opened the second front they really didn't have much of a chance. But it was a war where we [i.e. the English speaking countries] were almost* unquestionably the good guys. So we LIKE to think about it.)
* You might ask the opinion of the Poles about how much we were "good guys", but that's for leading them on and then not defending them after we'd promised to rather than for being too violent. Or you might ask the Finns. We were a lot less "good guys" than our history books claim. (But the point is that Germany had taken on over three times it's weight in armed might. It was depending on insane aggression to win...and that took it quite a ways, but it also eventually destroyed it.
When I looked I couldn't figure out how to write fixed length binary records to calculated locations. Something involving monads is all I could guess, but I still don't know what a monad is. Leibniz talked about monads as having no externals (no I/O ports?), but I'm not sure this is what OCaML meant.
It needs to be self documentation, because there's just about no documentation of Objective-C available...that isn't platform specific.
Sorry, but I've looked at learning Objective-C several times, and always bounced because of the truly pitiful documentation. Admittedly I didn't buy a book about how to use Objective-C on an iPhone (or whatever that was). I was interested in it as a programming language, and bounced when the documentation about that was so abominable. (I have the vague impression, e.g., that functions are only allowed to have one parameter. This *MUST* be wrong, but I couldn't find any documentation that would tell me that it was.)
Both actually. But this presumes that you're doing something appropriate to the language. I studied it for a month one time, but never felt impelled to go any further. It was designed before GUIs, and at it's peak of popularity was a mainframe end-user language. So no libraries (that I know of) have been written specifically for handling GUIs in APL. This means that you need to depend on Foreign Function Interfaces (FFI). "YUCK!", to quote me. I've never seen one of those that was cleanly implemented. (D to C is close, but only close. There's also Ada to C, which is nearly good. For the rest [of those I've tried to use], "YUCK!".)
OTOH, I've got to admit that I never tried to use APL's FFI. Or J's. Perhaps they're decent. But that's not the way I'd bet.
I think you're underestimating the complexity and sheer randomness of the thing being modeled.
If you think that "a real mathematician" could handle that problem, then you are probably a part of the problem. Some day some mathematician may handle it. Perhaps. And it will be a century before their theories are sufficiently tested that a good engineer would trust them. And they STILL would fall far short of a mathematical proof.
Einstein said it (paraphrase)"To the extent that mathematical theories apply to the physical world, they are uncertain. To the extent that they are certain, they do not apply to the physical world."
I would argue that the tax should be graduated, by how long you held the investment. The question in my mind is should it be a tax on the profits, or on the total investment. I lean towards a tax on the profits.
Say a tax on the profit of 40/sqrt(t) percent where t is measured in days. And any profits don't count as income for the purpose of income tax.
Has he solved the problem of evolutionary models tending to get stuck in cul-de-sacs then? Somehow I doubt it.
I believe that he could get fairly good results with that approach, but I think that their quality is limited. (How did his fund do over the last year?)
OTOH, given that most financial analysts don't know what they're doing, and an ape (gorilla?) in Chicago beat the average analyst one year, that's not saying much. I'd say that his fund has the advantage of a cheap financial analyst who's no worse than average. And that it's trying to beat a system with a very large random component. And unquantifiable externalities.
IOW, I don't think his system is likely to be very accurate. But neither is anybody else's. And his tea leaves are probably cheaper to read.
I don't think anyone has a good measure on how to include the externalities in a model. The only way to forecast their occurrence is to cause them. And that generally means either insider trading or causing a disaster.
You don't understand what a quant *is*. A quant is an employee. He doesn't make the decisions about policy. He tries to guess what's going to happen. His guesses are based on quantitative models, hence the term quant.
The quant probably cares deeply about robust models and statistical significance. This doesn't mean that his manager does, or even understands what the terms mean. The quant is, after all, basically a statistician. (He's not the salesman. That character has some different title.)
The quant probably isn't any more greedy than you are. Of COURSE he prefers a higher paying job. He's not a fool. But that's not his drive. OTOH, he's not in charge, either. And he's almost certainly not trying to sell anyone anything...well, no more than a programmer is. And he's probably no better at it than a programmer, either.
If you want to say "...The financial industry is completely amoral...." are you including the janitors? The secretaries? If not, then why are you including the quants? (If so, then you're just being ridiculous.)
Of course, this means you're assuming that your data approximates a normal curve, or that you have a two-way transform (bijective!) that maps it to such a curve.
Even if it had worked tactically, it was a strategic disaster. There's no way that the war in Iraq could have been a good thing for the country. OTOH, some people are getting rich, so it's been a good thing for them.
-
If treason weren't quite so narrowly defined by the constitution, then Bush would have knowingly committed treason. He knowingly acted in a way calculated to disadvantage the country. I'll grant that probably wasn't his purpose, but he had to know that that was one of the effects he was going to get. Even after WWII we had to maintain a LONG occupation. Well, that one was necessary...or at least it seems so from what I know. But it was VERY expensive, and not only in time and money, but also in terms of damage to the US social structures. There's no way that Iraq was worth that. Eventually we're going to be forced to withdraw, and just about everyone outside the US knows that. Certainly everyone with any political clout in Iraq. Until we do it's going to be a running sore, and there's no way a better outcome could have been expected. Things haven't gone as poorly as I expected, in fact, I'm rather heartened that it hasn't been worse (unless the news that I read is more censored than I know). But we don't have ANY real local support. Not everyone is opposed to us, and some are even pleased. But nobody is going to support us when they know that our stay is temporary. And they shouldn't.
-
Additionally, we haven't been acting in a way calculated to endear us to the local population. We've been quite high handed and brutal. Also capriciously violent and unpredictable. So even those who would like to be our friends don't trust us not to turn on them. This may make us slightly preferable to those they know would turn on them, but that's quite faint as praise goes.
-
Compare this to the way that we occupied either Germany or Japan. Well, the situations were different. In both those cases the countries had clearly initiated the violence, so everyone saw at least a modicum of justice in their being occupied afterwards. And the occupying troops didn't go out of their way to be unpleasant to people. (They weren't nearly as nice as the propaganda says they were, of course.) This is closer to the way that Hitler occupied France or Poland. (Sorry, that's the closest example that I know much about.) Nobody saw that as being just, so *of course* there were resistance fighters. Just as there are in Iraq. I'm not claiming that our troops act like the storm-troopers did, but at times it's much too close for any comfort at all.
The point is that ALL animals (and plants, etc.) are transgenic. There's nothing new or special about it. Various virus have been transporting genes cross-species for as long as we can tell (and there's no reason to think they started then).
What's new is doing it on purpose. (And we don't always use virus. Sometimes we use a golden shotgun...highly miniaturized, of course.)
Scientology is definitely not Gnosticism, and I resent the implication that it is. I'll grant some similarities, but then it's got more similarities to Freudian Psychology.
Gnosticism isn't authoritarian. It's true that most leaders of Gnostic groups have been...unreliable. That was usually intentional. Usually a part of the point was "Think for yourself, Schmuck!"
Actually, if a main point of the teaching isn't "Think for yourself!", then I deny that it can rightly be called Gnosticism. That is the central core of Gnosticism, and is why direct experience of the divine is so important. Scientology appears to be nearly the exact opposite. (And, of course, because of that it will have a certain symmetric relationship.)
FWIW, I've known people that quit Scientology. Granted this was several decades ago, and things may have changed. They didn't suffer any of the external harassment that is being claimed.
So perhaps what we hear about is the occasional exceptional case that is newsworthy? I don't assert that this is true, but it's worth considering. The news isn't an unbiased source of truth. If something isn't shocking or titillating, then you won't hear about it.
Yeah, I pretty much agree with you except for the bit about "clean coal". Unfortunately, that's an important difference because of where the govt. is choosing to invest money.
You forgot PL/I. Remember, IBM's favorite language. (It wasn't that bad, but much to complex for the time. Simpler than C++ though.) It, also, had pic formats. It borrowed them from COBOL.
I, personally, don't believe that it should be legal to use both copyright and patent to protect the same code. I consider doing so unethical. And I believe that it should be illegal. After you've released your code under an MIT or BSD (no attribution required) license, THEN I will consider that it might be ethical for you to have a patent. Until then, there is a prima facia case against it on ethical grounds. (Not, unfortunately, legal.)
Anything that CAN execute arbitrary ASM as native code CAN be optimal. Doesn't guarantee that it will be. Python can never be optimal, because of the overhead of the interpreter, but if most of the work happens in C coded libraries it can be quite fast enough.
I like Python too, but I sure didn't make the argument that you read. If I had I'd be coding my current application in Python instead of picking up C++. (Of course, if I had a large chunk of native binary that would do just what I want, then Python probably would be fast enough, but that's not an extant condition.)
P.S.: When I'm talking about executing arbitrary ASM code, I'm not talking about interpreting it, I'm talking about assembling it and executing it. Some languages that I know can do this are Forth and D. I believe that some dialects of C also have this built in. (I know they used to, but the last time I paid attention to this it was C running on a CP/M system. I don't believe that it's a part of standard C, but I could be wrong. When I said C could be hacked to execute binary I meant you could load arbitrary binary in from a file and then call it as a function. If you do it just right, you can make it work. Since this is only needed for very small pieces, like testing processor flags, etc., it's not a totally impossible thing to do. It's better to use a separate assembler routine, but it's not necessary, and I wanted to make a point that it was ultimately flexible in running self-modifying code. Also that such a thing was so complex and dangerous that there's probably never a decent reason to do it.)
FWIW, the analysis that I've seen says that it *is* a positive energy gain to use ethanol, so it would be possible without oil. It's just *quite* inefficient, so many other things would be better. (Possibly even methane.)
So. It's a bad idea. It's an expensive idea. It should be dropped immediately. But don't claim it couldn't be made to work.
P.S.: I've yet to see ANY evidence that clean coal can be made to work on any long term basis. Solar, wind, and nuclear are all viable choices as primary sources of energy. Some bio-diesel seems plausible. Cellulose conversion to ?? seems possible. (Works in the lab.) Coal....dubious. Coal to hydrocarbon is possible, but it's hardly carbon-free (i.e. no net emission of carbon). There are prototype "clean coal" plants in Scandinavia and, I believe, also Germany. I'm dubious about their ability to do long-term sequestration. I'm also dubious about the ability to assign costs when the sequestration fails in 20 or 30 years.
Actually, an optimal C program would be nearly as fast as any other application. But it might be horrendously complex. You can't really pick a problem that an optimal C program can't do nearly as quickly as any other program, esp. when you remember that in C you can hack binary code into running. The reasons to choose other languages are rather different. Easier to use and understand are right up there.
The obfuscated C competitions don't really give you any idea of what kind of gibberish an optimized C program can look like when it's handling a complex data structure. A result is that only mechanical optimization is common, and that's limited because the optimizer doesn't really understand what the program is doing. It would be interesting to see what a "evolutionary programming project" that operated on C source code would produce. I'd be willing to wager quite a bit that if it were optimizing anything complex, no human would understand the code in the first month. (Of course, it might take it quite a bit more than a month to write the code...) And the code STILL wouldn't be optimal!
Slashdot Mirror
Following several links from that I reach the conclusion that: /, etc. for the arithmetic operations. You need to call it with names like add:, sub:, mul:, div:, etc.
a) If you aren't developing for a Mac, or other Apple machine, you probably shouldn't bother with Objective-C, and
b) You can pass multiple parameters by defining a function which takes multiple named parameters, and using the names of the parameters each time you call. The names of the parameters will be considered a part of the name of the function.
c) And if you happen to define a numeric type, like, say, complex numbers, forget about using +, -, *,
It may have some advantages, but if so they were never displayed in the tutorials. (Actually, there appears to be a dialect of Objective C for the Mac [it wasn't stated that it was available elsewhere] which implements properties. I.e., getter and setter functions that can be addressed by the naked name of the variable. That's an improvement over C++ and Java. Not over many other languages, such as D and Python.)
To me it appears that Objective C/C++ has missed the boat. It appears to be falling behind various languages that are almost standing still. (I'm *not* impressed by it's way of allowing a function to accept multiple arguments. In Smalltalk that almost works, but Smalltalk has a much cleaner syntax than does Objective C. (And even in Smalltalk I find it bothersome.)
And programmers also very frequently know the results of what they're doing. If you want to blame the quants, go ahead. But be prepared to be bitten by the same tooth.
You KNOW that you didn't run enough tests, or that the program failed some of them. And you shipped it anyway. (And if you didn't know, you should have known.)
I've been spared that, because my company usually WAS my end user. Sometimes I released beta, but it was marked that way, and not used for anything crucial. Then it was fixed as requests came in. But there's no way the programmers in commercial houses can't KNOW that they haven't tested things enough. But if they do, they'll be outsold by the competition, because the customers can't tell the difference...until after, sometimes long after, they've bought it.
Sorry, it's been a few decades.
It didn't have any measurable chance of working, so playing "What if it did work?" is meaningless unless you're writing an alternate history novel. And then it means you'd do better to pick a different altered history. E.g., what if the US had supported the Geneva Conference on Viet Nam? That one has not only some plausibility, but also some potentially quite interesting consequences.
(OTOH, "What if the Germans won WWII?" has been done frequently, to profound artistic reaction, despite the fact that after Germany opened the second front they really didn't have much of a chance. But it was a war where we [i.e. the English speaking countries] were almost* unquestionably the good guys. So we LIKE to think about it.)
* You might ask the opinion of the Poles about how much we were "good guys", but that's for leading them on and then not defending them after we'd promised to rather than for being too violent. Or you might ask the Finns. We were a lot less "good guys" than our history books claim. (But the point is that Germany had taken on over three times it's weight in armed might. It was depending on insane aggression to win...and that took it quite a ways, but it also eventually destroyed it.
Because it isn't clear how to write a file?
When I looked I couldn't figure out how to write fixed length binary records to calculated locations. Something involving monads is all I could guess, but I still don't know what a monad is. Leibniz talked about monads as having no externals (no I/O ports?), but I'm not sure this is what OCaML meant.
cmucl is a Common Lisp. And I think he's got a Scheme dialect there too.
What's really strange is that gcc is considered as ONE compiler.
It needs to be self documentation, because there's just about no documentation of Objective-C available...that isn't platform specific.
Sorry, but I've looked at learning Objective-C several times, and always bounced because of the truly pitiful documentation. Admittedly I didn't buy a book about how to use Objective-C on an iPhone (or whatever that was). I was interested in it as a programming language, and bounced when the documentation about that was so abominable. (I have the vague impression, e.g., that functions are only allowed to have one parameter. This *MUST* be wrong, but I couldn't find any documentation that would tell me that it was.)
Both actually. But this presumes that you're doing something appropriate to the language. I studied it for a month one time, but never felt impelled to go any further. It was designed before GUIs, and at it's peak of popularity was a mainframe end-user language. So no libraries (that I know of) have been written specifically for handling GUIs in APL. This means that you need to depend on Foreign Function Interfaces (FFI). "YUCK!", to quote me. I've never seen one of those that was cleanly implemented. (D to C is close, but only close. There's also Ada to C, which is nearly good. For the rest [of those I've tried to use], "YUCK!".)
OTOH, I've got to admit that I never tried to use APL's FFI. Or J's. Perhaps they're decent. But that's not the way I'd bet.
I think you're underestimating the complexity and sheer randomness of the thing being modeled.
If you think that "a real mathematician" could handle that problem, then you are probably a part of the problem. Some day some mathematician may handle it. Perhaps. And it will be a century before their theories are sufficiently tested that a good engineer would trust them. And they STILL would fall far short of a mathematical proof.
Einstein said it (paraphrase)"To the extent that mathematical theories apply to the physical world, they are uncertain. To the extent that they are certain, they do not apply to the physical world."
I would argue that the tax should be graduated, by how long you held the investment. The question in my mind is should it be a tax on the profits, or on the total investment. I lean towards a tax on the profits.
Say a tax on the profit of 40/sqrt(t) percent where t is measured in days. And any profits don't count as income for the purpose of income tax.
Has he solved the problem of evolutionary models tending to get stuck in cul-de-sacs then? Somehow I doubt it.
I believe that he could get fairly good results with that approach, but I think that their quality is limited. (How did his fund do over the last year?)
OTOH, given that most financial analysts don't know what they're doing, and an ape (gorilla?) in Chicago beat the average analyst one year, that's not saying much. I'd say that his fund has the advantage of a cheap financial analyst who's no worse than average. And that it's trying to beat a system with a very large random component. And unquantifiable externalities.
IOW, I don't think his system is likely to be very accurate. But neither is anybody else's. And his tea leaves are probably cheaper to read.
I don't think anyone has a good measure on how to include the externalities in a model. The only way to forecast their occurrence is to cause them. And that generally means either insider trading or causing a disaster.
You don't understand what a quant *is*. A quant is an employee. He doesn't make the decisions about policy. He tries to guess what's going to happen. His guesses are based on quantitative models, hence the term quant.
The quant probably cares deeply about robust models and statistical significance. This doesn't mean that his manager does, or even understands what the terms mean. The quant is, after all, basically a statistician. (He's not the salesman. That character has some different title.)
The quant probably isn't any more greedy than you are. Of COURSE he prefers a higher paying job. He's not a fool. But that's not his drive. OTOH, he's not in charge, either. And he's almost certainly not trying to sell anyone anything...well, no more than a programmer is. And he's probably no better at it than a programmer, either.
If you want to say "...The financial industry is completely amoral. ..." are you including the janitors? The secretaries? If not, then why are you including the quants? (If so, then you're just being ridiculous.)
mean, variance, and skew?
Of course, this means you're assuming that your data approximates a normal curve, or that you have a two-way transform (bijective!) that maps it to such a curve.
Even if it had worked tactically, it was a strategic disaster. There's no way that the war in Iraq could have been a good thing for the country. OTOH, some people are getting rich, so it's been a good thing for them.
-
If treason weren't quite so narrowly defined by the constitution, then Bush would have knowingly committed treason. He knowingly acted in a way calculated to disadvantage the country. I'll grant that probably wasn't his purpose, but he had to know that that was one of the effects he was going to get. Even after WWII we had to maintain a LONG occupation. Well, that one was necessary...or at least it seems so from what I know. But it was VERY expensive, and not only in time and money, but also in terms of damage to the US social structures. There's no way that Iraq was worth that. Eventually we're going to be forced to withdraw, and just about everyone outside the US knows that. Certainly everyone with any political clout in Iraq. Until we do it's going to be a running sore, and there's no way a better outcome could have been expected. Things haven't gone as poorly as I expected, in fact, I'm rather heartened that it hasn't been worse (unless the news that I read is more censored than I know). But we don't have ANY real local support. Not everyone is opposed to us, and some are even pleased. But nobody is going to support us when they know that our stay is temporary. And they shouldn't.
-
Additionally, we haven't been acting in a way calculated to endear us to the local population. We've been quite high handed and brutal. Also capriciously violent and unpredictable. So even those who would like to be our friends don't trust us not to turn on them. This may make us slightly preferable to those they know would turn on them, but that's quite faint as praise goes.
-
Compare this to the way that we occupied either Germany or Japan. Well, the situations were different. In both those cases the countries had clearly initiated the violence, so everyone saw at least a modicum of justice in their being occupied afterwards. And the occupying troops didn't go out of their way to be unpleasant to people. (They weren't nearly as nice as the propaganda says they were, of course.) This is closer to the way that Hitler occupied France or Poland. (Sorry, that's the closest example that I know much about.) Nobody saw that as being just, so *of course* there were resistance fighters. Just as there are in Iraq. I'm not claiming that our troops act like the storm-troopers did, but at times it's much too close for any comfort at all.
I take it you never read the story.
The point is that ALL animals (and plants, etc.) are transgenic. There's nothing new or special about it. Various virus have been transporting genes cross-species for as long as we can tell (and there's no reason to think they started then).
What's new is doing it on purpose. (And we don't always use virus. Sometimes we use a golden shotgun...highly miniaturized, of course.)
Scientology is definitely not Gnosticism, and I resent the implication that it is. I'll grant some similarities, but then it's got more similarities to Freudian Psychology.
Gnosticism isn't authoritarian. It's true that most leaders of Gnostic groups have been...unreliable. That was usually intentional. Usually a part of the point was "Think for yourself, Schmuck!"
Actually, if a main point of the teaching isn't "Think for yourself!", then I deny that it can rightly be called Gnosticism. That is the central core of Gnosticism, and is why direct experience of the divine is so important. Scientology appears to be nearly the exact opposite. (And, of course, because of that it will have a certain symmetric relationship.)
FWIW, I've known people that quit Scientology. Granted this was several decades ago, and things may have changed. They didn't suffer any of the external harassment that is being claimed.
So perhaps what we hear about is the occasional exceptional case that is newsworthy? I don't assert that this is true, but it's worth considering. The news isn't an unbiased source of truth. If something isn't shocking or titillating, then you won't hear about it.
Yeah, I pretty much agree with you except for the bit about "clean coal". Unfortunately, that's an important difference because of where the govt. is choosing to invest money.
You forgot PL/I. Remember, IBM's favorite language. (It wasn't that bad, but much to complex for the time. Simpler than C++ though.) It, also, had pic formats. It borrowed them from COBOL.
You are forgetting that IBM has the necessary lawyers on the payroll, so it doesn't need to pay those retainers. They're already on salary.
You WERE explicit that it was a software patent.
I, personally, don't believe that it should be legal to use both copyright and patent to protect the same code. I consider doing so unethical. And I believe that it should be illegal. After you've released your code under an MIT or BSD (no attribution required) license, THEN I will consider that it might be ethical for you to have a patent. Until then, there is a prima facia case against it on ethical grounds. (Not, unfortunately, legal.)
Anything that CAN execute arbitrary ASM as native code CAN be optimal. Doesn't guarantee that it will be. Python can never be optimal, because of the overhead of the interpreter, but if most of the work happens in C coded libraries it can be quite fast enough.
I like Python too, but I sure didn't make the argument that you read. If I had I'd be coding my current application in Python instead of picking up C++. (Of course, if I had a large chunk of native binary that would do just what I want, then Python probably would be fast enough, but that's not an extant condition.)
P.S.: When I'm talking about executing arbitrary ASM code, I'm not talking about interpreting it, I'm talking about assembling it and executing it. Some languages that I know can do this are Forth and D. I believe that some dialects of C also have this built in. (I know they used to, but the last time I paid attention to this it was C running on a CP/M system. I don't believe that it's a part of standard C, but I could be wrong. When I said C could be hacked to execute binary I meant you could load arbitrary binary in from a file and then call it as a function. If you do it just right, you can make it work. Since this is only needed for very small pieces, like testing processor flags, etc., it's not a totally impossible thing to do. It's better to use a separate assembler routine, but it's not necessary, and I wanted to make a point that it was ultimately flexible in running self-modifying code. Also that such a thing was so complex and dangerous that there's probably never a decent reason to do it.)
FWIW, the analysis that I've seen says that it *is* a positive energy gain to use ethanol, so it would be possible without oil. It's just *quite* inefficient, so many other things would be better. (Possibly even methane.)
So. It's a bad idea. It's an expensive idea. It should be dropped immediately. But don't claim it couldn't be made to work.
P.S.: I've yet to see ANY evidence that clean coal can be made to work on any long term basis. Solar, wind, and nuclear are all viable choices as primary sources of energy. Some bio-diesel seems plausible. Cellulose conversion to ?? seems possible. (Works in the lab.) Coal....dubious. Coal to hydrocarbon is possible, but it's hardly carbon-free (i.e. no net emission of carbon). There are prototype "clean coal" plants in Scandinavia and, I believe, also Germany. I'm dubious about their ability to do long-term sequestration. I'm also dubious about the ability to assign costs when the sequestration fails in 20 or 30 years.
Actually, an optimal C program would be nearly as fast as any other application. But it might be horrendously complex. You can't really pick a problem that an optimal C program can't do nearly as quickly as any other program, esp. when you remember that in C you can hack binary code into running. The reasons to choose other languages are rather different. Easier to use and understand are right up there.
The obfuscated C competitions don't really give you any idea of what kind of gibberish an optimized C program can look like when it's handling a complex data structure. A result is that only mechanical optimization is common, and that's limited because the optimizer doesn't really understand what the program is doing. It would be interesting to see what a "evolutionary programming project" that operated on C source code would produce. I'd be willing to wager quite a bit that if it were optimizing anything complex, no human would understand the code in the first month. (Of course, it might take it quite a bit more than a month to write the code...) And the code STILL wouldn't be optimal!