random.org

[jeffjanzen]

it is my understanding that whenever a user requests a random sequence of events, linux generates pseudo-random numbers using an algorithm. (please correct me if i'm wrong.)
i'm not a programmer, but my guess is that it should be fairly simple and worthwhile to throw out the algorithm that's deceiving us and program linux to download true random numbers from random.org instead, so that users get true random sequences when they ask for them.
please read some of the literature on random.org to prevent the embarrassment of posting an ignorant reply.

[SSTwinrova]

Just to clarify, are you talking about the pregenerated files? If not, then you're assuming the computer has an active internet connection (while, even though probably more than half the time this would be true, could never be guaranteed to reach the 100% penetration mark that the offline "psuedo" generator provides).

[UbuWu]

In almost all cases, those pseudo random numbers are good enough for the goal that needs to be achieved.

[jeffjanzen]

"good enough"? the computer's hardware clock is "good enough" without daily ntp synchronization, but daily syncing is really the best way to go. i think random.org is the best way to go for random functions. it's just a little thing that could make linux better, the way i see it.
i should just learn the programming literature and rig it up myself, but i know nearly nothing of programming, and i'm afraid to break linux, because i like it so much and it's working pretty nicely for me now and i think some of you people that are reading this thread know enough to make this type of idea really fly in under an hour. (is that realistic at all?)
consider it an extra credit homework assignment. i don't want you guys to get bored...

[gord]

the hardware clock is just a seed as well, using a number from random.org to seed it would make the generated random numbers no more pseudo or valid than the regular numbers generated.

[jeffjanzen]

i'm not just talking about generating random numbers using a seed from a website, i'm talking about using true random numbers from random.org, in all forms (integers, sequences, other numbers, etc.) in linux applications, like randomly selecting which picture to use for a screensaver, or randomly assigning test subjects to groups for a scientific experiment.
fyi: "pseudo" means false.
i'm suggesting a movement away from pseudo-random numbers, toward using true random numbers. in most cases, our pseudo-random numbers are "valid" or "good enough," but the truth is that they're not actually random, and there are tasks (e.g. scientific experiments) that require true random values in order to be "valid" experiments. the numbers provided by random.org are truly random. please read:
http://www.random.org/essay.html
i suppose no one with the skills to implement such a function believes that true random numbers are necessary, or users should all go visit random.org themselves every time they want true random values. that kind of sucks, but i should have seen it coming...

[jeffjanzen]

i just discovered a wealth of linux software using random.org as a source for true random numbers. it feels good to know somebody shares my understanding of the occasional imperativeness of true random numbers.

EDIT: i was mistaken. a quick "random.org" query on unix.freshmeat.net yielded a few hundred applications that i hastily assumed used random.org in some way or another. truth: none of them do. (so often i'm addled by search engines....) i maintain that there are some situations which require true random numbers, and there are plenty who agree with me. there just aren't many who believe that these situations arise in linux, for linux users, and i'm starting to agree with them...

[kperkins]

i have just discovered a wealth of linux software using random.org as a source for true random numbers. it feels good to know somebody shares my understanding of the occasional imperativeness of true random numbers.

Care to share?

[Burke]

While the random number generator could definitely use some work (I don't know *how* to fix it, I just know it needs fixing), I think talking to random.org every single time you want a random number would bring your system to a grinding halt. rand(); in C++ takes a lot less time than I can determine, and grabbing data from some remote system, over even a fast connection still has relatively huge latency issues. Good idea in theory, but it would never work.

[kahping]

While the random number generator could definitely use some work (I don't know *how* to fix it, I just know it needs fixing), I think talking to random.org every single time you want a random number would bring your system to a grinding halt. rand(); in C++ takes a lot less time than I can determine, and grabbing data from some remote system, over even a fast connection still has relatively huge latency issues. Good idea in theory, but it would never work.

i think jeffjanzen is trying to say something like "replace the current pseudo-random number generator in Linux with the true random number generator algorithm used by random.org" or something to that effect.

i'd like to see that happen too, but i think the Linux developers have more important things to work on right now, so replacing something that's already "good enough" may not be a very high priority thing right now

kahping