Re: Crypto Mini-FAQ

From: Mok-Kong Shen (mok-kong.shen_at_t-online.de)
Date: 01/20/04 

Date: Tue, 20 Jan 2004 22:43:38 +0100

Lassi Hippeläinen wrote:
>

[snip]
> The only cryptoalgorithm that the academics have been able to prove
> secure is useless in real life. So we practising engineers have to live
> with the unproven kind, with all the mumbling, handwaving, and woodoo
> that comes with them. Maybe I should propose a conjecture: all good
> algorithms fall to Gödel's category "unable to prove either secure or
> insecure" ;-)

It seems intuitively clear that there can never be
a rigorous yet 'practically' applicable unit of
measure of 'strength' of crypto algorithms similar
to metre, second etc. in physics. So much of security
evaluation has to be more or less subjective. Actually
what faces engineers (I am referring to the fields like
mechanical engineering etc. that belong conventionally
to 'engineering') is not much essentially different,
for they also have 'unknowns'/unpredictables. What
they do is to apply in their designs a factor of safety,
which is somehow determined subjectively with reference
to experiences and certain empirical tests (full scale
or models). However, in crypto one could barely do
tests in the sense of engieering in my view. What makes
the whole even worse is the fact that there are secrets
of knowledge (apparently the agencies have certain
knowledge which they don't share with the public for
their good reasons), while this is not the case with
engineering, which is entirely public. I think that
these two factors essentially make one feel the
uncertainty in crypto uncomfortable in comparison to
the uncertainty in engineering which most people
generally take for granted.

M. K. Shen

Relevant Pages

  • Re: Projects
    ... cryptography for graduation? ... engineering, yes. ... Crypto typically requires PhD level mathematical knowledge. ...
    (sci.crypt)
  • Re: Is New Orleans finished ?
    ... It is not proper ... engineering. ... use as more wastes are stored.It will be monitored.Most importantly,it will be GUARDED,and secure. ... FAR safer and secure than where they stuff is NOW. ...
    (sci.electronics.design)
  • Re: Security in use of contants
    ... Neither Ruby nor Python are secure from reverse ... engineering, because all the source is hanging out visible for all to see! ...
    (comp.lang.ruby)