Wednesday, August 22, 2012

Truily Random Numbers


Gary Larsen
            
                                                                     SCIENCE
TRULY RANDOM NUMBERS

Having once toiled as a spreadsheet designer (engineer my card read), I can appreciate the significance of this- if not entirely understand it.

Random numbers are invaluable. They’re used in the encryption that makes online banking secure. Economists, physicists, pollsters, and casinos rely on them. Yet until recently, producing large sets of truly random digits has been hard to do.
[...]
The age of computers solved all of this, right? Wrong. The best that CPUs can generate are pseudo-random numbers, churned out by running a seed number through a complex algorithm, then running the solution through the same operation over and over again. However, anyone who uncovers the algorithm and seed can generate the same sequence of number.
[Here's the WTF part]
But now scientists at the Australian National University have introduced a technique for generating 5.7 billion truly random values per second. They do it by harnessing the fundamental uncertainty of the universe. Their technique measures quantum phenomena in a box completely devoid of photons, where ghostly virtual particles randomly burble in and out of existence 24/7. “God does not play dice,” Einstein famously quipped in response to evidence that randomness rules the cosmos. Luckily, he was wrong. [Full]


10 comments:

Jason in SD said...

They've been able to generate "true" random numbers using line noise for ages. Probably not 5.7 billion/sec of them.

Anonymous said...

My first thought upon reading this was "cool". My second, "does it have applicable value?" As an aside, quantum computers, if made practical, will make cracking encryption easy, as intractable problems will become solvable in polynomial time. That's the bad news. The good news is quantum encryption is now possible, where using the ability of quantum particles to change state when viewed to alert the message owner of eavesdropping. This has already been tested with cellphones. Amazing future we're building. Drummermanrick

Anonymous said...

If you live in a big city, the last 4 digits of telephone numbers "work". In high school, we calculated PI to a few decimal points by plotting thousands of these numbers to see whether they fell within the first quadrant of the unit circle. The ratio should be a quarter PI. Of course, that would be a lot easier now using a computer program with its pseudo random numbers and letting it crunch the numbers.

Freddie Sykes

BobG said...

Random and meaningless noise is much easier to produce; just ask a Liberal their opinion on anything.

DougM said...

Random is good.
Question you need to ask is: How random?
One coin flip is good enough for the NFL.
Pseudo-random generators are good enough for valid models.
Phone books are good enough for the lottery.
Any group of numbers is good enough for AGW models,
since any input produces a hockey-stick output.

Anonymous said...

As a crypto tech in the Navy, one piece of gear I worked on actually had a long-lived beta source to use as the initial key generator.

KellyFromMesquite

Anonymous said...

"They do it by harnessing the fundamental uncertainty of the universe."

The answer is always 42

thoR~

Kristophr said...

Random numbers are needed for secure cryptography, Drummermanrick.


You use them as seed values for keys used to hash things to be encrypted, and you can use a large book of random numbers to make 100% unbreakable one-use encryption pads.

For one use pads, both sides have the same pad. You use a part of the pad to encrypt the message, and delete the part used.

The person getting the massage used the same part of his pad to decrypt, and then also deletes that part.

Continue using said pad for messages until you run out of pad. Then you need to get a new pad and copy to replace them.

Anonymous said...

Sorry, thoR, but according to Thornton Melon the answer is "4?"

Freddie Sykes

Ole Phat Stu said...

How to get a fair result from tossing a biased coin :-

Toss it twice, if the result is HH or TT start again. The probabilities of HT and TH are equal and so can be used to get a fair result.

Post a Comment

Just type your name and post as anonymous if you don't have a Blogger profile.