[GreenKeys] FW: Link to BBC News Article: Colossus cracks codesonce more

[Craig Sawyers]

Hi all

> Colossus cracks codes once more
> By Mark Ward
> Technology Correspondent, BBC News website
>
> For the first time in more than 60 years a Colossus computer
> will be cracking codes at Bletchley Park.

I spent three days at the Heinz Nixdorf Museum (HNF) in Paderborn taking
care of the Lorenz SZ42 machine and assisting with transmitting the signals
for the cipher challenge (600W from the roof of the HNF).  There were thee
messages transmitted, with three different difficulties to crack: the
easiest had 7 wheels given and 5 to crack (2.2 x 10^7 possibities), medium 2
wheels given and 10 to crack (7.1 x 10^15 possibilities), and hard all
twelve weels to crack (1.6 x 10^19 possibilities).

The first complete break came from a computer and IT expert/radio ham in
Bonn called Joachim Scheuth.  He did a partial break of the difficult
message, and a complete break of the easiest on Thursday.  The easiest took
46 seconds processing time on his 1.6GHz machine.  He wrote his programme in
ADA, and somewhere (I don't have the link) he describes the attack he
adopted and makes available the code.

I don't think that Tony Sale at Bletchley tackled the easiest, because
reception conditions were lousy on the first broadcast day (Thurdsay).  But
they tackled the most difficult on Friday having got good copy, and made a
complete break with the rebuilt 1943 Colossus II in 3 hours 30 minutes.

The expected time for Colossus to break the easy challenge was around 10 to
15 minutes, which is interesting since it implies that a modern generation
PC is about a factor of 15 times faster than Colossus.  The comparison is
that Moore's law would indicate for 64 years a speed factor of 4 x 10^9!
Since the factor is only 10-15 times suggests that Colossus should not be
compared with a mere PC, but with a current generation supercomputer running
at Teraflops - or alternatively that Colossus was the equivalent of a
supercomputer in its day.  Of course, security services today do indeed use
massive, state of the art supercomputers for intelligence applications, so
nothing much has changed!

As an aside, the SZ42 had two settings - a wheel start position setting
(which corresponds to the challenges), and a bit pattern (which was fixed
and given for the challenge).  The way to understand the bit patterns is
that each wheel has a number of metal pegs which can be either set or
unset - and the peg operates the electrical contacts as the wheel rotates.
This can be thought of as a binary setting - so a 26 wheel has 2^26 possible
peg settings.  The total number of peg settings for the 12 wheels is about
10^150 and dramatically increases the difficulty of break.  Even in WW2
breaking these "wheel patterns" was testing, requiring luck with the
messages intercepts, ingenuity, and time.  A typical wheel pattern break
would take between 10 and 15 hours on a Colussus II.  By the end of the war,
there were 10 Colosus II machines at BP, running 24/7.

Craig