new idea for random number generation

Steve Kinney admin at pilobilus.net
Sat Aug 5 12:05:52 PDT 2017



On 08/05/2017 02:01 AM, Jan Dušátko wrote:

> And, the hardest one. What is exactly mean random number? Can choosing
> of numbers without any relations between them provide for us enought
> randomnes? How much of perfect random signal can be? This can lead to
> one crazy question - if the perfect random will be a small subset, can
> we be in risk use one of them?

That is a very tricky issue!  A true random number generator will
eventually output strings like "1234567890" that look anything but
random to a human.  So, does random mean "produced by an unpredictable
physical process," or does it mean "I do not see a pattern in the output
data"?  The latter answer is perilous, because our perceptual systems
seek patterns and happily impose them where none is present; faces seen
in cracked ceiling plaster, etc., are not "there" until we put them
there ourselves.

I see two answers:

A theoretical or abstract answer that says "As long as there is no way
to predict the placement or content of 'orderly' outputs in a stream of
data, that data is random."

A pragmatic answer says "Depending on the application a random number
generator is feeding into, it may be necessary to filter out strings
that present as 'too easy to guess' where and as they appear."

Does this kind of filtering make the remaining 'random' data less
random?  In the sense of "more predictable," yes it does.  But only
slightly, unless the pattern seeking filter is /very/ aggressive.

Calculating the impact of a filter that excludes strings like
"1234567890" from use as "random" data in applications like generating
cipher keys should be simple enough.  Increasing the length of the keys
to compensate for this bias, in the unlikely event that this is
considered worth doing, should be trivial in most cases.

> Lot of people do hard work understanding, what is it random. How we test
> random. How we detect random. And, how we can use random. We all enjoy
> their hard work.

Definitely.  Because of the close connection between the concept of
"random" and concrete physical phenomena that are easy to visualize and
manipulate in imagination, even I can understand and play with these ideas.

:o)



-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <http://lists.cpunks.org/pipermail/cypherpunks/attachments/20170805/435f6d20/attachment.sig>


More information about the cypherpunks mailing list