The problem is that these three functions are not robust over different populations. If they were, we would only observe the subset of patterns that can be produced by the specified higher function. And we dont! Most of the time there are systematic differences between cultures, genders, age.....
Most researchers just do some kind of EFA(explanatory factor analysis, hopefully using polychoric correlations) in one population on and on and pat themself on the shoulder when they can produce results that look similar to their theory. Nevermind that you could use the same data to produce results that fit many other theories. There is a reason that there never is a confirmatory step (cfa) after the explanatory.
Intelligence Tests as predictors are fine. After all they show your social standing, how well adapted and motivated you are and that you dont have any obvious mental problems and so on.
The problem is that if there was one type of intelligence certain patterns in solved problems/items would hold and they dont. You can ofcourse search for subpopulation of items/ppl where you can construct a metric scale but these also do not tend to be robust in replication. This point of view on intelligence is a positivist, empirical one and there are others like the definition of problem difficulty on kolmogorovs complexity. Problem is that reality doesnt give a fuck and we already know that there is no single difficulty/intelligence scale that can explain the observed patterns. more information here: http://en.wikipedia.org/wiki/Rasch_model http://www.amazon.com/Foundations-Measurement-Polynomial-Representations-Mathematics/dp/0486453146
From skimming their paper, it doesn't appear that they get any real speedup from their parallelism. This is apparent because they state, in the part about the millions of lines of code written in their language, that
These and other applications written in the source language are performance-competitive with established implementations on standard benchmarks
Translation: we didn't speed them up any, or at least not by enough that we care to share any number.
Amdahl's Law is difficult to overcome in auto-parallelising systems that rely on anything other than loop optimizations. Basically, in straight-line code, if you make 50% of the code run on multiple cores, you're only going to get at most a 2x improvement in speed. In practice, you won't even get that much (except in loops) due to added overhead. Bottom line: you can write papers about your wonderful auto-parallelizing technique, but when the rubber hits the road this is unlikely to lead to much performance improvement.
I do think that Amdahl's law is complete bullshit. There is no metaphysical part of the program that is not parallellizable . Parallell programming is, in contrast to concurrency, a performance optimization problem. It is the job of the programmer to come up with solutions, algorithms that perform better. After all, you can always do speculative evalution of later parts of the program if you have enough parallell ressources.
They have a Sigma 5 confidence level which is equivalent to one chance in a million that they're mistaken.
Can somebody please confirm/explain if the above quote is a correct interpretation and if so how.
Normally i would assume that the scientists use some kind of frequentist reasoning: Assume a H0. In this case the H0 would be no new particle. Run tons of tests and calculate how probable the results are under the H0. If the results are under a certain alpha level they throw away the H0 and assume the non H0 (ie. H1) as the new H0. Then calculate a confidence interval and report this.
Is this, in principle what happened? If so, then "one chance in a million that they're mistaken" is just plain wrong because the only probability/chance calculated was under the assumption that there is no new particle in which case there would be no new particle:)
In the first Iron Man film, Tony Stark is in a village in the Middle East and he kills a bunch of "bad guys" who are mixed in with a bunch of innocent civilians. He trivially distinguishes between his targets and the rest of the population.
This is bullshit. In real drone strikes, there is no guarantee that only "terrorists" are the victims. All the press reporting in the US takes the military at their word, and casualties are never identified as "collateral damage", i.e. innocent bystanders.
It's a real war, and there are always non-combatants who are killed and injured. Pretending this never happens may be good to keep support up at home, but it is a damned lie. Honesty is a better policy in the long run.
One of the reasons that Pakistan is not letting NATO resupply convoys go through it's territory is because of the toll taken by drone strikes. It is a huge issue with the Pakistan population. By not admitting to any civilian casualties in the US press, there can be no meaningful debate about how our policy is effecting US standing in the Middle East.
Personally, I think that the Pakistan government is not worth spit as an ally, and they are directly supporting our enemies. We would be better off if we cut most military aid because of their backstabbing behavior. Even so, the practical, ethical and political effects of our use of drones should be much more publicly debated, rather then being swept under the rub by what is effectively military propaganda.
In a real war soldiers of both sides are wearing uniforms.
In what world do you live in that publishing proprietary information on the internet doesn't get you fired? Google isn't in the habit of firing employees on a whim, so while might have thought he had approval, he clearly didn't from the people that mattered.
As for M$'s "research", most of it involes solving problems they created, or trying to clone Google's technologies. This is largely why you don't see any innovation coming out of their R&D, just Microsoft branded rehashes of existing and better things.
UMaple users can play MapleStory (using the MapleStory client software) without ever touching MapleStory's servers. UMaple then solicits "donations" that lead to enhanced privileges in the UMaple environment.
In this case some penalty does seem justified UMaple was after all making money from software written by MapleStory, without their permission
Why should the producers of MapleStory care if the infringer makes any money from it?
And what is the advantage of the start-button to metro. i have window 8 running on a laptop and i see(if i see it at all) as a glorified start menue. since the tiles are self actualizing it might also make a good screen saver. One advantage of metro is the availability of web apps and a centralized store. there are many js libraries that dont have a comparable.net equivalent (invoviz...). if you don't like it hit the windows key and poof gone.
I can recommend http://www.websharper.com/home F# and asp.net integration. they went open source a month ago. so far i am really impressed by it(formlets...).
Saw the link, watch the talk, pretty awesome. Language can be colorful at time, i don't know if it's typical in this setting.
Really liked the structure of his talk, and fact that it even goes into his state of mind when he worked on it really made the story telling much more interesting.
The armed forces they are deploying for thir conflict are speedboats and RPGs. Even if they are all gone in a few days, they can be replaced in a month. Wonder how much speedboats cost if you are buying them by the thousand? I would guess a lot less than the value of that oil trade.
and what country should replace those speedboats, since iran will long have seized to exist as a functioning state at that point?
Slashdot Mirror
The problem is that these three functions are not robust over different populations. If they were, we would only observe the subset of patterns that can be produced by the specified higher function.
And we dont!
Most of the time there are systematic differences between cultures, genders, age.....
Most researchers just do some kind of EFA(explanatory factor analysis, hopefully using polychoric correlations) in one population on and on and pat themself on the shoulder when they can produce results that look similar to their theory. Nevermind that you could use the same data to produce results that fit many other theories. There is a reason that there never is a confirmatory step (cfa) after the explanatory.
Intelligence Tests as predictors are fine. After all they show your social standing, how well adapted and motivated you are and that you dont have any obvious mental problems and so on.
The problem is that if there was one type of intelligence certain patterns in solved problems/items would hold and they dont.
You can ofcourse search for subpopulation of items/ppl where you can construct a metric scale but these also do not tend to be robust in replication. This point of view on intelligence is a positivist, empirical one and there are others like the definition of problem difficulty on kolmogorovs complexity. Problem is that reality doesnt give a fuck and we already know that there is no single difficulty/intelligence scale that can explain the observed patterns.
more information here:
http://en.wikipedia.org/wiki/Rasch_model
http://www.amazon.com/Foundations-Measurement-Polynomial-Representations-Mathematics/dp/0486453146
From skimming their paper, it doesn't appear that they get any real speedup from their parallelism. This is apparent because they state, in the part about the millions of lines of code written in their language, that
These and other applications written in the source language
are performance-competitive with established implementations
on standard benchmarks
Translation: we didn't speed them up any, or at least not by enough that we care to share any number.
Amdahl's Law is difficult to overcome in auto-parallelising systems that rely on anything other than loop optimizations. Basically, in straight-line code, if you make 50% of the code run on multiple cores, you're only going to get at most a 2x improvement in speed. In practice, you won't even get that much (except in loops) due to added overhead. Bottom line: you can write papers about your wonderful auto-parallelizing technique, but when the rubber hits the road this is unlikely to lead to much performance improvement.
I do think that Amdahl's law is complete bullshit. There is no metaphysical part of the program that is not parallellizable . Parallell programming is, in contrast to concurrency, a performance optimization problem. It is the job of the programmer to come up with solutions, algorithms that perform better. After all, you can always do speculative evalution of later parts of the program if you have enough parallell ressources.
http://www.amazon.com/Nokia-Lumia-920-Windows-Phone/dp/B00A2V7BA4/ref=zg_bs_2407747011_2
also posting to undo moderation.
posting to undo mderation.
Yeah and...
"640K ought to be enough for anybody." - Bill Gates, 1981
No matter how much space or bandwidth we have we will find a way to need more.
Source please.
That's an easy one:
"No."
Next question please.
They have a Sigma 5 confidence level which is equivalent to one chance in a million that they're mistaken.
Can somebody please confirm/explain if the above quote is a correct interpretation and if so how.
Normally i would assume that the scientists use some kind of frequentist reasoning:
Assume a H0. In this case the H0 would be no new particle.
Run tons of tests and calculate how probable the results are under the H0.
If the results are under a certain alpha level they throw away the H0 and assume the non H0 (ie. H1) as the new H0.
Then calculate a confidence interval and report this.
Is this, in principle what happened? :)
If so, then "one chance in a million that they're mistaken" is just plain wrong because the only probability/chance calculated was under the assumption that there is no new particle in which case there would be no new particle
Is licensing really done on a share of profit basis?`
Otherwise ist would make lots of sense to sue a high volume target that is being subsidized.
just use http://www.icsharpcode.net/OpenSource/SD/ or http://monodevelop.com/ .
http://en.wikipedia.org/wiki/Double_taxation#International_double_taxation_agreements
In the first Iron Man film, Tony Stark is in a village in the Middle East and he kills a bunch of "bad guys" who are mixed in with a bunch of innocent civilians. He trivially distinguishes between his targets and the rest of the population.
This is bullshit. In real drone strikes, there is no guarantee that only "terrorists" are the victims. All the press reporting in the US takes the military at their word, and casualties are never identified as "collateral damage", i.e. innocent bystanders.
It's a real war, and there are always non-combatants who are killed and injured. Pretending this never happens may be good to keep support up at home, but it is a damned lie. Honesty is a better policy in the long run.
One of the reasons that Pakistan is not letting NATO resupply convoys go through it's territory is because of the toll taken by drone strikes. It is a huge issue with the Pakistan population. By not admitting to any civilian casualties in the US press, there can be no meaningful debate about how our policy is effecting US standing in the Middle East.
Personally, I think that the Pakistan government is not worth spit as an ally, and they are directly supporting our enemies. We would be better off if we cut most military aid because of their backstabbing behavior. Even so, the practical, ethical and political effects of our use of drones should be much more publicly debated, rather then being swept under the rub by what is effectively military propaganda.
In a real war soldiers of both sides are wearing uniforms.
well, you have
http://cedet.sourceforge.net/
and the
http://ecb.sourceforge.net/ .
i dont think that there is much/if anything missing if you configure cedet correctly.
you're speaking like a typical windows user.
update? oh no, it will break things, try to detect my pirated software and make everything slower.
the typical linux user is more like:
upgrade? cool, new features, more stable software, better drivers.
and then the linux user wanders off and starts working on the dependecy graph.
In what world do you live in that publishing proprietary information on the internet doesn't get you fired? Google isn't in the habit of firing employees on a whim, so while might have thought he had approval, he clearly didn't from the people that mattered.
As for M$'s "research", most of it involes solving problems they created, or trying to clone Google's technologies. This is largely why you don't see any innovation coming out of their R&D, just Microsoft branded rehashes of existing and better things.
You got to be kidding me: http://research.microsoft.com/apps/dp/areas.aspx
UMaple users can play MapleStory (using the MapleStory client software) without ever touching MapleStory's servers. UMaple then solicits "donations" that lead to enhanced privileges in the UMaple environment.
In this case some penalty does seem justified
UMaple was after all making money from software written by MapleStory, without their permission
Why should the producers of MapleStory care if the infringer makes any money from it?
And what is the advantage of the start-button to metro. i have window 8 running on a laptop and i see(if i see it at all) as a glorified start menue. since the tiles are self actualizing it might also make a good screen saver. One advantage of metro is the availability of web apps and a centralized store. there are many js libraries that dont have a comparable .net equivalent (invoviz...). if you don't like it hit the windows key and poof gone.
Here is an older post about it: http://lambda-the-ultimate.org/node/4476
I can recommend http://www.websharper.com/home
F# and asp.net integration.
they went open source a month ago. so far i am really impressed by it(formlets...).
/absolutely agree.
Saw the link, watch the talk, pretty awesome. Language can be colorful at time, i don't know if it's typical in this setting.
Really liked the structure of his talk, and fact that it even goes into his state of mind when he worked on it really made the story telling much more interesting.
/agree
The armed forces they are deploying for thir conflict are speedboats and RPGs. Even if they are all gone in a few days, they can be replaced in a month. Wonder how much speedboats cost if you are buying them by the thousand? I would guess a lot less than the value of that oil trade.
and what country should replace those speedboats, since iran will long have seized to exist as a functioning state at that point?
So you are going to build your own facebook and then start creating applications for it?
Otherwise, having access to the source wont do you any good.
I cannot take them seriously anymore. Anyone to use them for business would be insane.
Yes right because Microsoft would never deprecate DCOM, Silverlight or VB6
can you point to any official statement that ms is going to pull silverlight?
If p is the chance, given no difference, of obtaining a result that is larger, what would you interpret (1-p) to mean?
not as the chance that there is a difference.