Embedding information in random bit sequences while maintaining certified randomness

[Jim Cheetham]

Ah, I hadn't remembered that was from Keith's talk :-)

I don't know where he got it from either, but there's an image on Wikimedia
that looks extremely similar ...
https://commons.wikimedia.org/wiki/File:Stereogramme_PI.jpg.

LCA has an annoying habit of running the 'current' conference in the same
namespace each year, and then archiving the old versions into a different
subdomain ... it would be better if they just published under "
lcaYYYY.linux.conf.au" and then redirected users to the current year ...

https://lca2017.linux.org.au/schedule/presentation/65/index.html

Our colleagues over at http://www.bitbabbler.org/what.html have a long
discussion about what statistical testing means to them, and from there I
get the idea that basically all these tests don't mean what we think they
mean ... they're only hunting for egregious failures.

Sadly I don't think I have the maths to describe why something that
obviously has structure (e.g. that stereogram) is potentially passing the
tests. I could throw up copies of dieharder & related tools, and try to
find data that they declared good, that I knew to be bad ... (like the
digits of various irrational numbers like pi itself?) but I'm not sure what
it would actually mean. The purpose of these tools isn't to say if a given
block of data is random, it's to make comments on repeated samples from an
RNG ...

-jim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ourshack.com/pipermail/discuss/attachments/20210831/985daee8/attachment.html>