I agree that the numbers are random, in the sense that they're subject to chance, but how confident are they that they know the sampling distribution? That is, can you use this method to generate a random sample a with uniform distribution, or a gamma distribution, or anything else you'd like to use random numbers for?
With quantum observation errors, I wonder if they're assuming the sampling distribution is normal, in which case they'd have to do some work to convert it to give the kind of output that rand() gives. Problems would likely show up in the tails of the distribution (near 0 and 1). TFA doesn't mention any of the statistical issues, only the physics ones.
iStat isn't necessary to cause the problem. Most people seem to be able to reproduce the issue by opening Photo Booth and trying to install/compile Boost from MacPorts (per these directions). FWIW, following these directions just crashed my 2011 15" MBP.
It seems to be an ATI graphics card issue, because some people report that they can get a stable machine by setting their graphics to integrated only with gfxCardStatus.
Sorta, but not quite a dupe. This post also includes speculation about what would happen if there were math errors in chips. The reference to Schneier's discovery about elliptic curve PRNG's was just to whet our appetite. But deliberate backdoors and ones created by mistake are two different things.
Slashdot Mirror
I agree that the numbers are random, in the sense that they're subject to chance, but how confident are they that they know the sampling distribution? That is, can you use this method to generate a random sample a with uniform distribution, or a gamma distribution, or anything else you'd like to use random numbers for?
With quantum observation errors, I wonder if they're assuming the sampling distribution is normal, in which case they'd have to do some work to convert it to give the kind of output that rand() gives. Problems would likely show up in the tails of the distribution (near 0 and 1). TFA doesn't mention any of the statistical issues, only the physics ones.
iStat isn't necessary to cause the problem. Most people seem to be able to reproduce the issue by opening Photo Booth and trying to install/compile Boost from MacPorts (per these directions). FWIW, following these directions just crashed my 2011 15" MBP.
It seems to be an ATI graphics card issue, because some people report that they can get a stable machine by setting their graphics to integrated only with gfxCardStatus.
Sorta, but not quite a dupe. This post also includes speculation about what would happen if there were math errors in chips. The reference to Schneier's discovery about elliptic curve PRNG's was just to whet our appetite. But deliberate backdoors and ones created by mistake are two different things.
TFA is just a summary of an article yesterday in the NYT: http://www.nytimes.com/2007/11/17/technology/17code.html?ref=technology