Join 3,376 readers in helping fund MetaFilter (Hide)


Genuine random numbers, generated by radioactive decay
February 9, 2009 7:46 PM   Subscribe

HotBits is an Internet resource that brings genuine random numbers, generated by a process fundamentally governed by the inherent uncertainty in the quantum mechanical laws of nature, directly to your computer in a variety of forms. HotBits are generated by timing successive pairs of radioactive decays detected by a Geiger-Müller tube interfaced to a computer. (Warning: random sounds.)
posted by parudox (41 comments total) 8 users marked this as a favorite

 
I like. Pseudo random numbers, eat shit.
posted by unSane at 7:50 PM on February 9, 2009 [2 favorites]


xkcd://221
posted by mikelieman at 7:53 PM on February 9, 2009


And they're doing it using https? Good for them!

Of course, if you're using the random data for cryptography or other security-related applications, you can't be certain I'm not squirreling away a copy. But I'm not, really.

I think I like this person.
posted by Lemurrhea at 7:56 PM on February 9, 2009


This is one of those things that I'm certain I would appreciate were I all sciency and stuff.
posted by oddman at 8:04 PM on February 9, 2009 [1 favorite]


Genuine random numbers are so much better than those that are indistinguishable from random numbers.
posted by norabarnacl3 at 8:08 PM on February 9, 2009 [2 favorites]


Genuine random numbers are indeed better than those that are supposed to be indistinguishable from random numbers, but may or may not be.
posted by unSane at 8:13 PM on February 9, 2009 [1 favorite]


Randomness without radioactivity.
posted by Joe Beese at 8:14 PM on February 9, 2009 [1 favorite]


Genuine random numbers are so much better than those that are indistinguishable from random numbers.

For cryptographic purposes, yes. If you know the pseudorandom number generator algorithm used, you can predict one pseudorandom number from the one before. But you wouldn't use this for cryptographic purposes; you'd capture your own entropy and avoid trusting someone else.
posted by musicinmybrain at 8:15 PM on February 9, 2009 [3 favorites]


If I use this, the next time a nat 20 comes up I will finally be void of suspicion.
posted by incompressible at 8:23 PM on February 9, 2009


If you think you might need some random numbers when you aren't connected to the Internet, I highly recommend the RAND Corporation's magnum opus, A Million Random Digits with 100,000 Normal Deviates. RAND reprinted it a few years ago, and now it can be yours for only $81.
posted by grouse at 8:31 PM on February 9, 2009


There's a Reaktor soundscape generator that uses white noise from an unplugged sound card input jack as a random source. Is that a good enough random source for cryptography? I think it's the background noise of the stars, like primitive radio astronomy.

Come to think of it, wouldn't any old wav file of ambient street noise be good enough for a random source? Just treat the individual samples as values. I have no idea of mathematical cryptography, I'm just thinking out loud here.
posted by fleetmouse at 8:36 PM on February 9, 2009


grouse: "If you think you might need some random numbers when you aren't connected to the Internet, I highly recommend the RAND Corporation's magnum opus, A Million Random Digits with 100,000 Normal Deviates. RAND reprinted it a few years ago, and now it can be yours for only $81."

Fuck those warmongers. You can get a stick of 5 casino-grade dice for $15.
posted by Joe Beese at 8:37 PM on February 9, 2009


I think more things should be sold in sticks.
posted by JHarris at 8:48 PM on February 9, 2009 [2 favorites]


Last year when I was a CS student and found out about the pseudo-random thing, I developed a solution within five minutes.

A small, possibly clear box (because who wouldn't want to see this? It's awesome!) housing a small robotic hand and two, or three, or ten 20-sided dice. When you press a button on the device, the hand picks up the dice, shakes them, and lets them fall. Besides looking really cool, it should, in theory, produce random numbers, as long as the dice themselves aren't weighted. And it would look really cool.

Never understood why someone didn't create such a device. I guess it's the one shortcoming of silicon--it only does one thing with no variation whatsoever. To solve some problems, you have to literally think outside the box. It's just kind of funny to me, because this radioactive decay idea is basically a more mature version of something I sketched in a notebook a year ago while I was bored in a freshman-level class.
posted by DMan at 9:01 PM on February 9, 2009 [1 favorite]


int getRandomNumber()
{
   return 4; // chosen by fair dice roll.
             // guaranteed to be random.
}

posted by grumblebee at 9:01 PM on February 9, 2009 [4 favorites]


Photon measuring for random numbers. In the past 1.5 years, they've served up 5/4ths of a Terabyte of randomness. Not bad.
posted by Lemurrhea at 9:14 PM on February 9, 2009


Weird, I discovered the existence of A Million Random Digits with 100,000 Normal Deviates just this afternoon. Is the English translation faithful to the original J.L. Borges?
posted by tepidmonkey at 9:14 PM on February 9, 2009 [4 favorites]


The usual Amazon.com customer review hijinx for the book of random numbers
posted by Rhomboid at 9:23 PM on February 9, 2009 [4 favorites]


I'll make sure to use this for any studies I run, just to be able to cite it in a write-up: "Random numbers for stimulus location were generated by radioactive decay."
posted by parudox at 9:43 PM on February 9, 2009 [4 favorites]


If you want a digital edition to go with your dead-tree edition of RAND's 1M Random Digits, you can always buy (or download) a 600MB CD-ROM of well-vetted random numbers.

A few years ago (probably 10+ at this point), I saw a little gadget that you could connect to your PC's parallel port, which acted as a true random-number generator. I think it used resistor thermal noise, or some other quantum effect, to do its magic. But it was sold for the truly paranoid to use as a source of perfect random entropy for priming PRNGs, primarily for cryptography applications. I never saw one in the wild and never heard of it again (although admittedly, until today I'd never really looked). I think it might have come from this company.

My understanding is that today, most modern processors have TRNGs on-die, capable of producing the entropy necessary for encryption keys and other cryptographic tasks (and possibly small-scale Monte Carlo simulations if you really needed that level of randomness), but that the software support is somewhat lacking, because PRNGs are "good enough" for almost everything.
posted by Kadin2048 at 10:07 PM on February 9, 2009


Oh, great. I don't even know what it means, but because it's there, I now have to sift through 2048 bytes of "random" hexadecimal data so I can see the face of the Cylon God.

People call me crazy, but I say you're the crazy ones. THEY HAVE A PLAN, PEOPLE!
posted by ford and the prefects at 10:22 PM on February 9, 2009


Check out the "100 heads" coin flipping robot.

It's just kind of funny to me, because this radioactive decay idea is basically a more mature version of something I sketched in a notebook a year ago while I was bored in a freshman-level class.

I would say coin flips and die rolls are really deterministic chaotic processes that are very hard to model, rather than nondeterministic random processes. It's not unthinkable that a high speed camera and sufficiently advanced software will be able to predict die rolls one day. However, if you manage to predict the timing of radioactive decay precisely, you have just rewritten how the universe works.
posted by benzenedream at 10:40 PM on February 9, 2009


I was going to call rubbish on this, until I saw the URL. It's not often that you can help shape an industry and eventually have a science-fiction book include your swiss lab as a final showdown location...
posted by jkaczor at 10:49 PM on February 9, 2009


At MeFi, we love randomness.
posted by Zed at 11:28 PM on February 9, 2009


I read this too fast. I thought it was about random hobbits.
posted by ActingTheGoat at 11:37 PM on February 9, 2009


I will never use this, or tell anyone of my knowledge, but I am stoked.
posted by cmoj at 11:52 PM on February 9, 2009


You know what they say, clicking that link killed the cat.
posted by fatllama at 12:45 AM on February 10, 2009 [2 favorites]


[...]the software support is somewhat lacking, because PRNGs are "good enough" for almost everything.

Someone should tell the guy who wrote the docs for Python's random module, to wit:

random.shuffle(x[, random])

Shuffle the sequence x in place. [...] Note that for even rather small len(x), the total number of permutations of x is larger than the period of most random number generators; this implies that most permutations of a long sequence can never be generated.

A true source of randomness would not present this problem.
posted by JHarris at 1:34 AM on February 10, 2009 [1 favorite]


(Er, not to mention negating an entire category of certificate attacks.)
posted by JHarris at 1:37 AM on February 10, 2009


"There's a Reaktor soundscape generator that uses white noise from an unplugged sound card input jack as a random source. Is that a good enough random source for cryptography? I think it's the background noise of the stars, like primitive radio astronomy."
If it's picking up radio noise, whether astrophysical in origin or not, I could conceivably sit outside your office and generate a signal that makes your generator sufficiently non-random to be a problem. I don't know the details of how this soundscape generator is supposed to work, but it sounds comparatively risky to the truly paranoid.

It's harder to do this with a Geiger counter. You'd have to have a high output gamma source (other radiation wouldn't trigger the counter or wouldn't have the penetrating power to do it from an obscured location) and it's pretty difficult to come up with a way to generate large numbers of gamma rays in such a controlled fashion, I think.
posted by edd at 3:06 AM on February 10, 2009


I just use Amazon Mechanical Turk.
posted by oulipian at 3:15 AM on February 10, 2009 [1 favorite]


Body : 5753D6B46A577FFAC7577242C55ADB583D2C539AB913FE5951EF2B35A82B3CDD
27B5E76E984E76C042B6B0004C463BF50F9767A8464FE5653186684741C3AFE2
56E7E8E0A3A8543778BD5FFB7334B7D0C2BE82A010E297E274FAC13630C53052
75705672B8662917BCA74C218A7DEED230DA59EC7C8A403A87361575887A51E4

Quickness: 1590D1A4EF17D52CF1020F3FFB84E8101229BB60563468A943EE5F0A99A3F173
09592940650BCFF2A624205F6945DAEED52604B53C82847D7FCAD832B2CCB4F8
BB3EF79FE93EA1C4A303215BDD7998DE23D363D46769E2575388CA1CC87C2254
392E4DC7D3D3F3CBD360A6E11934D98B4A18FD58D50EBF841D6ABBBDDDE0929F

Strength: 3F701807BD91E898E0B1A62368DE94C6E0712E70E3637E71A15BC396483F72CF
764EE6509A718BB58AA25C62ED4ED571B46700B224B675B0D53871536CEDA7B2
3B073F0383CABE490B6FE2FE50142D2E2A291BD607ABF5B7DD87C0524D4E1D79
253389CF6EA3BC5110FC1E8DC6EEB40032EFBF89F82E5CAC21B2B31A4D7D949C

Intelligence: 4CBCA1DD9AE5D4CFB51BE98447112DE35A2859D1BB283D40C34FFA62B6DAA102
C2C850931427D8130EFC4CD6456F45FA94733BE64F932603FEF292EC7740DC4D
4ED9343E2C311DE120860F19DE566AEB7D715279F683E429466C0F25957F36A9
74C065E66E411EF1BD4FF0B459B50AB638D25CC5517B846F24F44AA3B530F04B

Willpower: B3C8BD544F0F3B747A4D13423D5DC2B058C4B0952E7E0D03F6A25349DDF8AB03
CDB3822C3B96B904D8D802027CD97473A88FA439E428D4497D6CB402877487AC
1A98800309D550F0373942F33C0AC65D07B47607ABD603D78FD53E7918A2FB2A
1D75E31E2509D221DE87DE7E1D001205524F30E02174A759AC20216427FD2B23


Charisma: 6EA208E18A03FBF28D55B3FEA4A986A5FFDAB4DEB3C16C29CF1D0BE6F5CBA218
1FBEAFA7DCCA52CB89DB2E4DFE98B1F7929B5B5441194EDFFAB9F7EE7302C456
DD0EB65684D43D3402184861C022CCD6389C954589DA276B349E1FB5F6F6B20F
BDD24B8E96699610BE7610A456CB010716D407D38EC722B2A62A371E3A158891

Essence: 1DE020D87FF4D1B0E0F642E9081EC30DD40C0CEEA4AA560615D3A5C2D1787FB2
B0B11E9304DABB4DC7C6835500D4DBBFAE20EDD8DB0A81898D36EB527BBA3BDF
C77F58D56BB2F04AF06C23EEBB3DFE61B61E3D9CEDA1F8878717AEE10D899422
816EA2DF2376EB325179B07B59A4510FA4140629DECA6F354F14AD45A2F831FD

Magic: 4833FC33364AE738B563572B543F444DDC134E0B5DBF33D8929B4D87FD2E9767
D6862E6498A4952630F811F4CB8849F9B13272A1E9F91FAEEF7D7A49BA212B10
E7B19C6C906A262899DF63BA7F3C8B1DC685EC12D66D7A87025178B22D9F63CB
770E2A34E7DC05908032679C53FEFA9F741E69478C20506E51A7798B72B7E1B9

Karma Points: 9

Suck it metahumans!
posted by Potomac Avenue at 5:12 AM on February 10, 2009 [1 favorite]


...it should, in theory, produce random numbers, as long as the dice themselves aren't weighted.

No, it should, in practice, produce random numbers. In theory they are perfectly predictable from newtonian physics. Radioactive decays, on the other hand, are in fact random in theory.
posted by DU at 5:41 AM on February 10, 2009 [3 favorites]


Actually that's not strictly true. Quantum mechanics doesn't strictly require the probabilistic approach. It can be deterministic as long as you allow it to be nonlocal. Although it's perfectly reasonable to say that the experimenter can't determine the hidden variables that underly the determinism, so practically speaking it's still random and I'd suspect just as good for cryptographic purposes.
See here.
posted by edd at 6:06 AM on February 10, 2009 [1 favorite]


Thanks DU, you are of course correct.

The idea of taking radio noise from an open sound card input is pretty creative, although it's a bit too easy to defeat. I wonder what other strange solutions exist for the problem....obviously, not everyone has access to a Geiger counter/radioactive decay.
posted by DMan at 7:11 AM on February 10, 2009


It would be relatively simple to construct a cloud chamber and sit a web cam in front of it. It'd be harder to get the software to convert trails of subatomic particles to random numbers and more difficult to do transformations on your inputs to get suitable uniform or gaussian distributions, but it'd be relatively straightforward to construct the physical side of it.

The bit rate would be kind of low too - it only solves the Geiger counter problem, not the radioactive source problem.
posted by edd at 7:19 AM on February 10, 2009


Note that for even rather small len(x), the total number of permutations of x is larger than the period of most random number generators; this implies that most permutations of a long sequence can never be generated.

Oh, this is bad. This is really bad.
posted by grouse at 8:09 AM on February 10, 2009


not everyone has access to a Geiger counter/radioactive decay

We need a smoke detector RNG.
posted by benzenedream at 9:24 AM on February 10, 2009


DMan: Last year when I was a CS student and found out about the pseudo-random thing, I developed a solution within five minutes.

A small, possibly clear box (because who wouldn't want to see this? It's awesome!) housing a small robotic hand and two, or three, or ten 20-sided dice. When you press a button on the device, the hand picks up the dice, shakes them, and lets them fall.


Some problems with dice.
posted by Pronoiac at 9:51 AM on February 10, 2009


Random post.
posted by Pronoiac at 12:54 PM on February 10, 2009


Kadin2048: If you want a digital edition to go with your dead-tree edition of RAND's 1M Random Digits, you can always buy (or download) a 600MB CD-ROM of well-vetted random numbers.

So. Did anyone else look at that, wonder why they don't offer a zip or compressed tarball of it, then kick themselves?

And is it just me, or does it seem like the PRNG for the latest couple commenters is broken?
posted by Pronoiac at 2:37 PM on February 10, 2009 [4 favorites]


« Older Every Swear Word from The Sopranos...  |  "This was a symbolic killing,"... Newer »


This thread has been archived and is closed to new comments