But we aren't talking about total stopping distance, we are talking about the distance between the two cars. That is a linear relationship. Again, assuming equal and constant braking acceleration, we have:
t = v/a, so time to stop is linear wrt v.
In the time it takes the following driver to react, the first car acquires a velocity of vdiff = a t_react wrt the second car, and a small reduction in the distance between the cars (t_react is small relative to t.) Thereafter, we have d = vdiff t, so d is linear wrt t, and therefore v.
It is a linear relationship, but your distances are off by a factor of 3 for typical traffic. The rule of thumb is two seconds between cars ( 100*1000/60/60*2 = 55m at 100kph.)
Sadly, because many roads are way overloaded, people cut this to 1 second or so, which works with good driver attentiveness, good visibility, and good road conditions, provided nothing unexpected happens. When something bad happens (e.g. a jack-knifing t-t, a fog pocket, ice, etc,) you get to see spectacular multi-vehicle pile-ups -- sometimes 10+ cars are involved before the information can propogate to the following traffic.
Yes, but that doesn't have much to do with anything.
In the military, following unlawful orders is a no-no. You don't get to do it and offload your responsibility on the officer issuing the order. Stupid, but legal, orders must generally be obeyed: if ordered to take a hill with a stupid plan, you still have to try to do it, you just report that enemy fire makes the plan unworkable. I.e. you follow orders, but orders don't require suicide. At worst, the issue is sorted out later at a court martial.
School is a bit different - the Supremes have held students don't even get full 1st amendment rights while in school. The teachers are assumed to be the responsible decision makers, and students are not liable for acts they commit at the teachers' order.
Using a browser doesn't rise to the level of rights for students. The teacher doesn't want it, it doesn't happen. End of story.
Well, actually, the story continues, but the story is only that the teacher is an idiot.
1. You have a crappy codebase. 2. New builds therefore exhibit new bugs. 3. So tons of reg tests are done on builds to ensure behaviour didn't change. 4. Then the release cycle goes from hours to days then to months. 5. And the programmers start moving functionality from code to data, cos data is more mutable according to The Rules. 6. And things get even worse cos there's more pointless code in the system. 7. And you add more managers to control the whole process.
Eventually, if you are lucky, senior management sees it has a clusterfuck on its hands, and hires someone to fix it. Sadly, a lot of the time, management hires the wrong person, and the whole thing just dies.
There is no statistic to determine fraud, almost by definition. The various statistical tests look for improbable departures from expectations, and the fraudster tries to modify the data in a way that doesn't look improbable.
Given we have few datasets of fraudulent vs non-fraudulent numbers, it is hard to generate hard numbers. Instead, we look at tests the fraudsters didn't consider or understand, and these tests usually show such extreme numbers that any statistician would assume the data was manipulated. For example:
1. Faked biology data (several known examples) - means look good, but higher order stats are way outside a normal distribution. Luckily, you can repeat the experiments, and see the repeats don't show the reported results.
2. Faked accounting data (tons of examples.) Most fakers make really basic mistakes. E.g. around 27%? of financial numbers should begin with 1, faked data usually has the wrong leading number distribution. Again, forensic accountants dig here and usualy hit paydirt.
3. Image manipulation. Again, the manipulator gets the first order stats right, but leaves a mess in terms of higher order stats (local vs global noise.)
I'm not making an argument for the existence of God. I'm simply saying that those who say "there is no God" have gone beyond the ability of science since a negative cannot be proved. Do you dispute that?
I'd dispute that. Science doesn't waste its time with Gods, FSMs, pixies, demons, or anything else. It tries to divide statements into true, false, unknown, and pointless. Your God stuff is in the pointless category - it seems to have no effect on anything. By Occam's razor, we consider pointless a subset of false, but would be happy to move your claim into into either unknown or true if you could point to any interesting phenomenon we should look at.
We don't deny there are an infinite number of unprovable statements, it's just that treating them all equally doesn't advance science.
We are not in a declared war with any state. North Korea?
It's kind of hard to declare a war against an ambiguous enemy. Huh? We declared a war on poverty, and then a war on drugs.
Enemy combatants are identified by behavior, not by uniform or flag. Since they are a militia of no government (and if they were, of no government we are at war with, since we have not declared war with any government that remains) these enemy combatants caught in acts of aggression are mere criminals and are not in fact prisoners of war. So, we pick person X, see that we are not at war with his government, declare him a "mere criminal", and incarcerate him without POW status or the consent of his goverment? Even if he donned a uniform to fight as a partisan?
Wow, I guess all those Polish and French resistance guys deserved what they got.
About the only the good place to research futher is Wall Street. There might be some minor players like the DoD, etc.
Almost nothing has been published, but a few people at all the big firms know the issues: data is 2d (it has an arrival time and an effective time,) restatements are common, we need multi-dimensional time-travel to explain variables, etc.
Having spent the last few years designing and building temporal databases, I'm with you on this one. You just can't do it with SQL (that's your assembly language, not the programmer environment.)
Most people don't need, want, or understand the stuff that temporal models can provide, but those that do tend to react with dropped jaws and "OMFG"s when you show them a system that does this stuff.
I lost a compute farm at 1am last week (only people who noticed were the two PHBs working past midnight.) Called support, and got voicemail. Took three plus hours to resolve, overall cost was merely in the high 5 figure range.
You keep my machines running, you get paid well. I don't care if you're watching youporn or playing games. You let a multi-million dollar facility go dark and don't even notice, there will be tears at year's end.
Re:Have some patience, we'll run across them... ev
on
The Fermi Paradox is Back
·
· Score: 2, Interesting
That is not true. It's not that the conditions and chemical constitution of the environment need to be the same, it's the fact that their needs to be a very low probability event (or set of events) occur for the first "living" cell to result from some arbitrary water-based reaction somewhere in the planet, giving us a cell that has at least basic reproduction and respiratory (energy converting) capabilities. Evolution cannot aid this first cell: there are none before it. It has to come as a result of a single "miracle moment" where the necessary compounds for a connected cell wall, nucleus, DNA..etc all form at the same time AND at the same small point in space, albeit at a much smaller degree of complexity compared to living cells today.
Wow. No serious scientist has proposed life starting by a cell miraculously springing into existence with no prior evolution involved.
Most of the pre-biotic soup theories involve dilute mixtures of animo aids, peptides, sugars, polymers, etc, that replicate as a group. No DNA or similar is involved, there are no cell walls, little or no respiratory capabilities. These features all evolve incrementally and independantly over time. As Darwin noted, the "first life" might have been a salty, slightly greasy, tidal pool.
Wow. You understand more than most game designers out there.
That's the sort of commentary I always hope to hear from an interview candidate (pity I'm in an entirely different field.) Then we do the 45 minutes of in-depth discussion.
Easy answer: it was made to protect children, not to ban thought crimes.
Googled text from 2002: WASHINGTON (CNN) -- The U.S. Supreme Court Tuesday struck down a 6-year-old law that prohibits the distribution and possession of virtual child pornography that appears to -- but does not -- depict real children.
The law had banned a range of techniques -- including computer-generated images and the use of youthful-looking adults -- which were designed to convey the impression of minors engaging in sexually explicit conduct.
The 6-3 ruling says the law violates the First Amendment guarantee of freedom of speech. The decision hands a major setback to the Justice Department and the majority of Congress in their legislative efforts to fight child pornography.
The supremes' actually bitch-slapped the govt pretty hard in this one. Their older ruling explicitly said computer-generated stuff wasn't child porn. The govt selectively quoted the supremes' earlier ruling to imply computer-generated stuff was illegal. Hilarity ensued.
That is not what the ninth amendment says, nor does it reflect the intent of the writers.
The enumeration in the Constitution, of certain rights, shall not be construed to deny or disparage others retained by the people.
Pretty simple: sounds like unnamed rights are not to denied or weakened because other rights were explicitly enumerated in the Constitution.
Lets see if that was the intent of the writers. Madison:
It has been objected also against a Bill of Rights, that, by enumerating particular exceptions to the grant of power, it would disparage those rights which were not placed in that enumeration; and it might follow by implication, that those rights which were not singled out, were intended to be assigned into the hands of the General Government, and were consequently insecure. This is one of the most plausible arguments I have ever heard against the admission of a bill of rights into this system; but, I conceive, that it may be guarded against. I have attempted it, as gentlemen may see by turning to the last clause of the fourth resolution.
Hmm, he agrees with the orignal poster, not you.
Hamilton?
The exceptions here or elsewhere in the constitution, made in favor of particular rights, shall not be so construed as to diminish the just importance of other rights retained by the people; or as to enlarge the powers delegated by the constitution; but either as actual limitations of such powers, or as inserted merely for greater caution.
He's probably somewhat right: Wall Street would like these kind of models, but the problem is that they tend to appear implicitly in the structure of the firms. I.e. even with the model, profits would probably not increase.
You can think of a big Wall Street firm as a federation of desks (i.e. model implementors,) with each desk in turn being a federation of smaller desks or strategies. These desks focus on specific areas of expertise (e.g. FX volatility, Mortgage defaults.) They tend to lay off the risk of the bits they don't care about by hedging (i.e. making deals) with other desks in the firm (so, for example, a commodities desk trades with the FX desk to remove their risk of foreign exchange movements.) The net result of this is that the system is largely self-balancing: by management giving the desks risk and capital constraints, the desks trade with each other to optimize their own profits and thus a fairly optimal global solution emerges.
This global solution is driven not by a global model, but rather ny many niche models (each with lots of specific business expertise) interacting efficiently.
Time is often helpful in determining causation. The problem is that observations occur after causes, but if we miss underlying causes, spurious causality appears (e.g. ice cream sales are high at noon, at 4 p.m. we see a peak in heatstroke cases - naively, eating ice cream causes heatstroke.)
I spent some time working on weaker measures of direction: X and Y are measured simultaneously, but we compute Dominate(X,Y) and Dominate(Y,X) -- i.e. does X explain Y better than Y explains X? This, of course, assumes non-linear relationships. Made a decent amount of money on Wall Street before the other quant shops caught up, though:)
Gack. So by "direction" your source means the sign of co-effecient. And thus "strength" is unsigned. No disagreement here then.
But, when talking about correlation and causation, "direction" is used to describe causal effects: see the second use of "direction" in the Wiki article you cited.
...indicates the strength and direction of a linear relationship
It assuredly does not indicate direction in any way: Corr(X,Y) exactly equals Corr(Y,X). You can trivially prove this by renaming the variables in the equation and then rearranging the multiplications to get the original equation.
I'm not too sure on the memory usage of other scripting languages. LSL was designed to be compiled in bytecode (like python) and then executed in virtual machines. The maximum amount of memory a script can use (including it's bytecode) on SL is 13kb. Are you sure the alternatives would be able todo the same LSL can do within the same memory constraints?
Anything I can look at to compare with?
Pretty sure. Most stack-based VMs (Smalltalk-80, Forth, Python, etc) are approximately equivalent (with a factor of 3) in terms of memory and CPU efficiency.
One nice thing you get with the bytecode VM/many user processes approach is that you can write a better memory allocator than general C: assign each user process a fixed block of RAM (say 16K,) and repoint malloc to just grab a block from this space, repoint free to do nothing. Just kill any user process that goes over bounds. It's a fast mempool approach that avoids fragmentation on long running servers.
Do the same thing for CPU: just kill any user process that eats more than 1M bytecode executions. You're now well on your way to a serious server running several hundred thousand untrusted user processes.
Exactly. The energy conversion efficiency of lungs+muscles is about 3% versus around 30% for an internal combustion engine (for fairly obvious reasons: much higher temp/pressure in engines means more efficiency, people are people not high efficiency energy->work convertors.)
Slashdot Mirror
But we aren't talking about total stopping distance, we are talking about the distance between the two cars. That is a linear relationship. Again, assuming equal and constant braking acceleration, we have:
t = v/a, so time to stop is linear wrt v.
In the time it takes the following driver to react, the first car acquires a velocity of vdiff = a t_react wrt the second car, and a small reduction in the distance between the cars (t_react is small relative to t.) Thereafter, we have d = vdiff t, so d is linear wrt t, and therefore v.
It is a linear relationship, but your distances are off by a factor of 3 for typical traffic. The rule of thumb is two seconds between cars ( 100*1000/60/60*2 = 55m at 100kph.)
Sadly, because many roads are way overloaded, people cut this to 1 second or so, which works with good driver attentiveness, good visibility, and good road conditions, provided nothing unexpected happens. When something bad happens (e.g. a jack-knifing t-t, a fog pocket, ice, etc,) you get to see spectacular multi-vehicle pile-ups -- sometimes 10+ cars are involved before the information can propogate to the following traffic.
Yes, but that doesn't have much to do with anything.
In the military, following unlawful orders is a no-no. You don't get to do it and offload your responsibility on the officer issuing the order. Stupid, but legal, orders must generally be obeyed: if ordered to take a hill with a stupid plan, you still have to try to do it, you just report that enemy fire makes the plan unworkable. I.e. you follow orders, but orders don't require suicide. At worst, the issue is sorted out later at a court martial.
School is a bit different - the Supremes have held students don't even get full 1st amendment rights while in school. The teachers are assumed to be the responsible decision makers, and students are not liable for acts they commit at the teachers' order.
Using a browser doesn't rise to the level of rights for students. The teacher doesn't want it, it doesn't happen. End of story.
Well, actually, the story continues, but the story is only that the teacher is an idiot.
Aargh! I've seen this anti-pattern so many times!
1. You have a crappy codebase.
2. New builds therefore exhibit new bugs.
3. So tons of reg tests are done on builds to ensure behaviour didn't change.
4. Then the release cycle goes from hours to days then to months.
5. And the programmers start moving functionality from code to data, cos data is more mutable according to The Rules.
6. And things get even worse cos there's more pointless code in the system.
7. And you add more managers to control the whole process.
Eventually, if you are lucky, senior management sees it has a clusterfuck on its hands, and hires someone to fix it. Sadly, a lot of the time, management hires the wrong person, and the whole thing just dies.
There is no statistic to determine fraud, almost by definition. The various statistical tests look for improbable departures from expectations, and the fraudster tries to modify the data in a way that doesn't look improbable.
Given we have few datasets of fraudulent vs non-fraudulent numbers, it is hard to generate hard numbers. Instead, we look at tests the fraudsters didn't consider or understand, and these tests usually show such extreme numbers that any statistician would assume the data was manipulated. For example:
1. Faked biology data (several known examples) - means look good, but higher order stats are way outside a normal distribution. Luckily, you can repeat the experiments, and see the repeats don't show the reported results.
2. Faked accounting data (tons of examples.) Most fakers make really basic mistakes. E.g. around 27%? of financial numbers should begin with 1, faked data usually has the wrong leading number distribution. Again, forensic accountants dig here and usualy hit paydirt.
3. Image manipulation. Again, the manipulator gets the first order stats right, but leaves a mess in terms of higher order stats (local vs global noise.)
I'm not making an argument for the existence of God. I'm simply saying that those who say "there is no God" have gone beyond the ability of science since a negative cannot be proved. Do you dispute that?
I'd dispute that. Science doesn't waste its time with Gods, FSMs, pixies, demons, or anything else. It tries to divide statements into true, false, unknown, and pointless. Your God stuff is in the pointless category - it seems to have no effect on anything. By Occam's razor, we consider pointless a subset of false, but would be happy to move your claim into into either unknown or true if you could point to any interesting phenomenon we should look at.
We don't deny there are an infinite number of unprovable statements, it's just that treating them all equally doesn't advance science.
That's pretty funny. It has no relationship to reality, of course. Probably worth +5 informative on slashdot, though.
Um, the professor "pretended to marvel" according to the story. Of course he knew what a side-rule was.
We are not in a declared war with any state.
North Korea?
It's kind of hard to declare a war against an ambiguous enemy.
Huh? We declared a war on poverty, and then a war on drugs.
Enemy combatants are identified by behavior, not by uniform or flag. Since they are a militia of no government (and if they were, of no government we are at war with, since we have not declared war with any government that remains) these enemy combatants caught in acts of aggression are mere criminals and are not in fact prisoners of war.
So, we pick person X, see that we are not at war with his government, declare him a "mere criminal", and incarcerate him without POW status or the consent of his goverment? Even if he donned a uniform to fight as a partisan?
Wow, I guess all those Polish and French resistance guys deserved what they got.
About the only the good place to research futher is Wall Street. There might be some minor players like the DoD, etc.
Almost nothing has been published, but a few people at all the big firms know the issues: data is 2d (it has an arrival time and an effective time,) restatements are common, we need multi-dimensional time-travel to explain variables, etc.
Want to give it a shot?
Having spent the last few years designing and building temporal databases, I'm with you on this one. You just can't do it with SQL (that's your assembly language, not the programmer environment.)
Most people don't need, want, or understand the stuff that temporal models can provide, but those that do tend to react with dropped jaws and "OMFG"s when you show them a system that does this stuff.
Odd number? Ha, count the blades on all the fan in your residence: primes (3,5,7,11, etc) Anything else invites lower frequency noise.
I know, you may find a cheap 4 blade fan, but you won't find a 24 blade PC cooling fan.
I'm a PHB, and totally agree with you.
I lost a compute farm at 1am last week (only people who noticed were the two PHBs working past midnight.) Called support, and got voicemail. Took three plus hours to resolve, overall cost was merely in the high 5 figure range.
You keep my machines running, you get paid well. I don't care if you're watching youporn or playing games. You let a multi-million dollar facility go dark and don't even notice, there will be tears at year's end.
That is not true. It's not that the conditions and chemical constitution of the environment need to be the same, it's the fact that their needs to be a very low probability event (or set of events) occur for the first "living" cell to result from some arbitrary water-based reaction somewhere in the planet, giving us a cell that has at least basic reproduction and respiratory (energy converting) capabilities. Evolution cannot aid this first cell: there are none before it. It has to come as a result of a single "miracle moment" where the necessary compounds for a connected cell wall, nucleus, DNA..etc all form at the same time AND at the same small point in space, albeit at a much smaller degree of complexity compared to living cells today.
Wow. No serious scientist has proposed life starting by a cell miraculously springing into existence with no prior evolution involved.
Most of the pre-biotic soup theories involve dilute mixtures of animo aids, peptides, sugars, polymers, etc, that replicate as a group. No DNA or similar is involved, there are no cell walls, little or no respiratory capabilities. These features all evolve incrementally and independantly over time. As Darwin noted, the "first life" might have been a salty, slightly greasy, tidal pool.
Wow. You understand more than most game designers out there.
That's the sort of commentary I always hope to hear from an interview candidate (pity I'm in an entirely different field.) Then we do the 45 minutes of in-depth discussion.
Easy answer: it was made to protect children, not to ban thought crimes.
Googled text from 2002:
WASHINGTON (CNN) -- The U.S. Supreme Court Tuesday struck down a 6-year-old law that prohibits the distribution and possession of virtual child pornography that appears to -- but does not -- depict real children.
The law had banned a range of techniques -- including computer-generated images and the use of youthful-looking adults -- which were designed to convey the impression of minors engaging in sexually explicit conduct.
The 6-3 ruling says the law violates the First Amendment guarantee of freedom of speech. The decision hands a major setback to the Justice Department and the majority of Congress in their legislative efforts to fight child pornography.
The supremes' actually bitch-slapped the govt pretty hard in this one. Their older ruling explicitly said computer-generated stuff wasn't child porn. The govt selectively quoted the supremes' earlier ruling to imply computer-generated stuff was illegal. Hilarity ensued.
That is not what the ninth amendment says, nor does it reflect the intent of the writers.
The enumeration in the Constitution, of certain rights, shall not be construed to deny or disparage others retained by the people.
Pretty simple: sounds like unnamed rights are not to denied or weakened because other rights were explicitly enumerated in the Constitution.
Lets see if that was the intent of the writers. Madison:
It has been objected also against a Bill of Rights, that, by enumerating particular exceptions to the grant of power, it would disparage those rights which were not placed in that enumeration; and it might follow by implication, that those rights which were not singled out, were intended to be assigned into the hands of the General Government, and were consequently insecure. This is one of the most plausible arguments I have ever heard against the admission of a bill of rights into this system; but, I conceive, that it may be guarded against. I have attempted it, as gentlemen may see by turning to the last clause of the fourth resolution.
Hmm, he agrees with the orignal poster, not you.
Hamilton?
The exceptions here or elsewhere in the constitution, made in favor of particular rights, shall not be so construed as to diminish the just importance of other rights retained by the people; or as to enlarge the powers delegated by the constitution; but either as actual limitations of such powers, or as inserted merely for greater caution.
Wow, he agrees with the original poster too.
He's probably somewhat right: Wall Street would like these kind of models, but the problem is that they tend to appear implicitly in the structure of the firms. I.e. even with the model, profits would probably not increase.
You can think of a big Wall Street firm as a federation of desks (i.e. model implementors,) with each desk in turn being a federation of smaller desks or strategies. These desks focus on specific areas of expertise (e.g. FX volatility, Mortgage defaults.) They tend to lay off the risk of the bits they don't care about by hedging (i.e. making deals) with other desks in the firm (so, for example, a commodities desk trades with the FX desk to remove their risk of foreign exchange movements.) The net result of this is that the system is largely self-balancing: by management giving the desks risk and capital constraints, the desks trade with each other to optimize their own profits and thus a fairly optimal global solution emerges.
This global solution is driven not by a global model, but rather ny many niche models (each with lots of specific business expertise) interacting efficiently.
Time is often helpful in determining causation. The problem is that observations occur after causes, but if we miss underlying causes, spurious causality appears (e.g. ice cream sales are high at noon, at 4 p.m. we see a peak in heatstroke cases - naively, eating ice cream causes heatstroke.)
:)
I spent some time working on weaker measures of direction: X and Y are measured simultaneously, but we compute Dominate(X,Y) and Dominate(Y,X) -- i.e. does X explain Y better than Y explains X? This, of course, assumes non-linear relationships. Made a decent amount of money on Wall Street before the other quant shops caught up, though
Gack. So by "direction" your source means the sign of co-effecient. And thus "strength" is unsigned. No disagreement here then.
But, when talking about correlation and causation, "direction" is used to describe causal effects: see the second use of "direction" in the Wiki article you cited.
As usual, Wiki sucks.
, and vice versa.
Which means there is NO directionality.
...indicates the strength and direction of a linear relationship
It assuredly does not indicate direction in any way: Corr(X,Y) exactly equals Corr(Y,X). You can trivially prove this by renaming the variables in the equation and then rearranging the multiplications to get the original equation.
I'm not too sure on the memory usage of other scripting languages. LSL was designed to be compiled in bytecode (like python) and then executed in virtual machines. The maximum amount of memory a script can use (including it's bytecode) on SL is 13kb. Are you sure the alternatives would be able todo the same LSL can do within the same memory constraints?
Anything I can look at to compare with?
Pretty sure. Most stack-based VMs (Smalltalk-80, Forth, Python, etc) are approximately equivalent (with a factor of 3) in terms of memory and CPU efficiency.
One nice thing you get with the bytecode VM/many user processes approach is that you can write a better memory allocator than general C: assign each user process a fixed block of RAM (say 16K,) and repoint malloc to just grab a block from this space, repoint free to do nothing. Just kill any user process that goes over bounds. It's a fast mempool approach that avoids fragmentation on long running servers.
Do the same thing for CPU: just kill any user process that eats more than 1M bytecode executions. You're now well on your way to a serious server running several hundred thousand untrusted user processes.
Right. The scripting language seems like a typical home-made C clone: Lua, Python, etc, would do just as well with less cost to all involved.
I was hoping to see a langauge that really captured the concepts of virtual worlds and actors. LSL ain't it.
Exactly. The energy conversion efficiency of lungs+muscles is about 3% versus around 30% for an internal combustion engine (for fairly obvious reasons: much higher temp/pressure in engines means more efficiency, people are people not high efficiency energy->work convertors.)