German amateur code breaker defeats Colossus

An amateur cryptographer from Germany has beaten Colossus, the world’s first programmable digital computer, in a code-breaking challenge. The original machine was developed at Bletchley Park to crack encrypted German messages during World War II. After years of painstaking restoration work a recreation of machine returned to …

[Tom Chiverton]

Um...

"allowing him to uncrack the message"

Uncrack ? That's the same as encrypting, right ?

[Simon Martin]

Programming Dentists? A pedant enquires....

ADA is the American Dental Association

I think the language used was Ada (after Ada Lovelace)

[lordasb]

During the war...

Great to see this in the news last night on the TV and on the reg :-)

i visited about 4 weeks ago and its very impressive... nicely layed out and man in charge is very friendly and chatty.

well worth a visit!!!!

[Anonymous Coward]

Bull5hit!

'Colossus was a single function device whose speed at breaking codes rivals that of modern PCs'

Really?

In fact Colossus is roughly as fast as a 'Colossus EMULATOR' running on a knackered old P2 (no other spec given - how fast, how much RAM, etc)

So, first of all, it's as fast as a geriatric PC running an EMULATOR which may be well written - or not.

Now see how well it does against a modern PC, say a nice shiny new fully loaded 3ghz Core2Duo based device running dedicated code breaking software that knows all about WW2 Axis encryption systems.

That would be a fair comparison with a 'modern PC'.

THEN try saying the it's speed at code cracking rivals that of a modern PC!

[Wile E. Veteran]

Tricky language?

Ada tricky? It's largely just Pascal on steroids. Oracle's PL/SQL is just a thinly-disguised subset of Ada.

Now, if he'd used APL....

[Dazed and Confused]

Which prime minister?

Was it Churchill?

Although Colossus was created to crack the Lorenz code of the German High Command. I though that they didn't start to dismantle Bletchley Park till after the end of the war in the Far East. So would it not have been the post war labour government that ordered the destruction?

[Anonymous Coward]

Nope!

'Colossus was a single function device whose speed at breaking codes rivals that of modern PCs'

ORLY?

> His 1.4 GHz PC needed 46 seconds to decipher the message

http://www.heise-security.co.uk/news/99090

PWND!

[Peter Gathercole]

APL ...

... it's all greek to me!

[Jeff]

Ada Programming is not Tricky

I suspect that coding the algorithm that did the code breaking was tricky. Programming it in Ada was not. We have kids come in as summer interns with no Ada experience who are doing decent programming within a few days. Like most of the modern languages, it may take years to really understand everything but it is certainly no more tricky than anything else out there.

[Anonymous Coward]

He probably just guessed the encryption key...

It might have been something obvious like "don't mention the war" ;-)

[amanfromMars]

Future Perfect CodeXXXX.

Is ADHD a tangent/manifestation of Ada's Enigmatic Code?

In the Upper Reaches/dDeeper Growth of the Jungian Freud Masters and Johnson Amazon Jungle ......IT is Virgin Territory for Nymph and Satyr alike ...... and a QuITe Heavenly Store of Immaculate Source? :-)

Can you tell me that it would not be so .... and why you would choose to think that so?

[Mark Fenton]

@Peter Gathercole

Didn't you mean "It's all GEEK to me" ?

:)

[Andrew Moore]

Like for like???

Did the programmer first intercept the signal via radio first like Bletchley Park???

[Anonymous Coward]

@Mark Fenton

Actually, G(r)eek... (the 'R' is silent...)

:))

[Vulpes Vulpes]

The coded message ...

I can exclusively reveal, was, "Don't tell him, Pike!"

[Karl Lattimer]

RE: Bull5hit!

You should really sort yourself out, first of all, a von neuman machine (a modern computer) cannot perform the kind of operations colossus did as colossus was a dedicated non-deterministic number cruncher of extra-ordinary design. In fact I believe it was designed by a postal worker, the same postal worker that designed the first automated sorting machine, the implementation was based loosely on a turing bombe. The only machine today that can realistically process data in the same way is the Cell BE.

Emulating colossus is therefore not possible, the universal turing machine emulators can process a colossus program, however it is massively slower than running on a dedicated turing machine (as are all turing programs).

Using the method he did, he managed to crack it quickly, well done to him, I doubt it resembled the original method of cracking as von neuman machines didn't actually exist when colossus was built, and further investigation into the codes since the war would have given some more efficient methods of cracking.

Either way, one inaccuracy that has been reported over and over, Churchill did not order the destruction of all 10 Colossi in fact, only 9 were destroyed. A tenth is apparently at GCHQ to this day, possibly still in operation... Who knows!

The reason one machine was kept? I'd imagine just in case someone started using Enigma/Lorenz or a similarly constructed cipher, the machine would still be available if required.

[Liam Johnson]

@Andrew Moore Like for like??

According to the Heise report, yes, he did receive the signal via radio, but he was a lot closer and was not limited to using WWII equipment.

[Paul Crawford]

@Nope

Well is says:

"Bonn resident Joachim Schüth (call-sign: DL2KCD) managed to stabilise and read the very noisy radio transmission of the Cipher text at 12:00 UTC and decipher it by 14:00"

So that to me is 2 hours, even if the actual run on the PC was 45 seconds.

Now considering even that, 45 seconds (now) versus 3h35m (then) we have a speed improvemnt of 287 times (tempted to say "286" here!) which is not anything like 50+ years of "Moore's Law".

What would you have said 10 years ago for the relative PC performance?

[Anonymous Coward]

Re: Karl Lattimer

The engineer largely responsible for Colossus was indeed a Post Office man, the late Tommy Flowers:

http://en.wikipedia.org/wiki/Tommy_Flowers

I remember seeing a documentary on Bletchley Park which was made not long before the great man died in 1998. I was delighted to see that despite being in 90s, he was using an IBM PC and he was still clearly in possession of all his faculties. If I get to that age I just hope I still have as much get up and go as he had!

[Tony]

@Karl

I think that you will find that the "Postal Worker" was one of the senior engineers for the GPO telecommunications division (what became BT) by the name of Tommy Flowers.

I understand that he offered the plans to the War Office before outbreak of hostilities - they couldn't see a need for such a device. Despite that, he built one himself and spent approx £2000 in the process (a 3 bed house at the time would have been about £350).

After the (don't mention the war) a greatful Government awarded him the princely sum of £75 for his sterling efforts; although he also later received an MBE. (Just shows that they are as consistent now as they were then)

According to some sources he also worked on the development of ERNIE; thanks to him, I've won about £600!

[Bob H]

@Bull5hit!

I'm not normally pedantic, but I am tired of people using the incorrect unit capitalisation for frequencies.

IT IS "GHz" people not Ghz, ghz, GhZ, GHZ or any of the other permutations I see on a daily basis!

[Anonymous Coward]

Re: Tony

According to the (occasionally reliable) Wikipedia, Tommy Flowers got £1000 for his work on the Colossus. Still not enough to cover the money he spent out of his own pocket, but considerably more than the £75 you cited. On reading some more, I was interested to find out that at least ten Colossus machines were built, two of which were used well into the 1950s.

[PunkTiger]

Code cracking...

It's a Good Thing™ that this competition wasn't held in the States; because we all know the DMCA automatically makes such actions illegal.

[Anonymous Coward]

So what?

Hardly surprising that someone with earlier access to the message and a modern computer could decode it first.

Perhaps if the chap in Germany had been required to build his own computer from scratch without any modern (i.e. silicon/transistor) equipment and then derive the necessary algorithms to decode it, like the people at Bletchley did, I'd be a bit more impressed.

[Frank Bough]

Paul Crawford

"Now considering even that, 45 seconds (now) versus 3h35m (then) we have a speed improvemnt of 287 times (tempted to say "286" here!) which is not anything like 50+ years of "Moore's Law"."

...you seem to neglect that Schüth's CPU is likely smaller than his little fingernail and about one billionth the cost of Colossus in real terms.

[Ian]

@Frank Bough

"...you seem to neglect that Schüth's CPU is likely smaller than his little fingernail and about one billionth the cost of Colossus in real terms."

So all the billions (trillions?) of <insert your currency of choice here> spent by numerous academic and commercial organisations globally over the last 50 years developing and improving all aspects of computer technology to the point where Herr Schüth can have his teeny-tiny powerful CPU in his own personal computer doesn't count? You only count the purchase price of the machine itself? But the much smaller amount of money spent on developing colossus does count?

Seems like an unfair comparison.

My take on Paul Crawford's comment was simply that Colossus was way ahead of of its time (and perhaps that you can't extrapolate Moore's law too far back in time before Moore initially stated it?)

[Derek Hellam]

The last machines

I believe that they keep the last machine and the fact that they had broken the codes secret until the 1970's, because various pro-Soviet states and various Soviet departments still used machines similar to the enigma long after the war ended. So we where still reading their codes during the cold war. Correct me if I'm wrong

[Anonymous Coward]

@ Bob H

Gigs to you!

[Anonymous Coward]

Oh really?

1. C isn't a programming language, it's shit shorthand for assembler. So is c++. They both made it because they were free, and kiddie coders prefer playing to engineering. Unix is here because it was free, not good. It's now reliable because of billions of users testing it, rather than any inate brilliance. Straustrup failed when he designed c++. Java was the language he might have designed, though c# would have been better, and vb.net better still, but you can only push children so far, and he knew it.

2. A programme written in Ada is likely to be easier to understand than any other language. It was deisgned to be easy to compile check out semantic mistakes, and easy to read.

3. Wouldn't he be a cryptanalyst if he broke the codes? Cryptographers make codes, Cryptanalysts break them.

4. Churchill didn't order the destruction of the Collossi, because post war we gave away the rotor machines to friendly nations, advertised as uncrackable, so we could spy on them.

[Anonymous Coward]

@Oh really?

> Java was the language he might have designed, though c# would have been better, and vb.net better still...

Better for Billy Gates and his band of madcap marketers, maybe, but honest folks (i.e. those outside Microsoft) who'd prefer software to work properly before it's recruited into the profit jihad that Microsoft's unlawful abuse of its monopoly (just ask the US DoJ or the European Commission) has become, wouldn't agree.

[Paul]

C shorthand for assembler?

"C shorthand for assembler".. well, great! real programmers program in machine code, not assembler, since they know all the op-codes by heart in decimal, hex and binary - the latter is essential since you toggle your code into the machine using the switches on the front panel.

However, we real programmers get lazy with age and like to use assembly language, or even C. C++ is, however, for quiche eating script kiddies!

[Jamie]

Welcome to the Machine!

See pic.

http://news.bbc.co.uk/nol/shared/spl/hi/pop_ups/07/technology_enl_1195122923/img/1.jpg

Nah still faster than my ZX81...

[David Wilkinson]

given what passes for real programming today (javascript)- Ada is difficult.

The old standard was that its not a real language unless you can write on OS with it.

Furthermore you weren't a "real" programmer unless you were proficient in at least one "real" programming language.

Now that scripting languages no matter how product specific/feature limited is a "real" programming language.

By that standard Ada is difficult...

[Anonymous Coward]

Bletchley Park

I highly recommend you all visit Bletchley Park - I have been to the Burma Railway, Normandy Beaches, Ardennes, etc..., and nowhere was as thrilling as being at Bletchley Park and seeing where the war was really fought and won.

I met the bloke rebuilding the Colossus when I went there about 5 years ago and the whole scene was amazingly cool.

Highly recommended.

[Steve]

Shorthand?

C shorthand for assembler? Sure you're not thinking of BLISS ? :)

At least we no longer use RTL/2. All the worst features of C and Pascal in one language...

I'll second (third) the recommendations for Bletchley Park. Well worth a visit, and it will take you at least a whole afternoon.

[A J Stiles]

Ada?

I always thought the language Ada was famous for not including a way to read a single character from an input stream (such as a keystroke).

Or maybe they've fixed that now.

[Tony Rogers]

ENIGMA Settings

This cracking of a code is a bit confusing for me.

As I read and understood, the machine was developed to ascertain the settings

on the Enigma machine which were independently levers that were changed

to alter the code. When the new settings were evaluated the code could be read from the Enigma machine.

[Steve]

@Tony Rogers

It was the Bombe that was developed to decode Enigma messages, by identifying the settings. Colossus was not used for Enigma decoding, but for cracking other codes (Lorenz, I think).

[Anonymous Coward]

@ Paul

"the latter is essential since you toggle your code into the machine using the switches on the front panel"

Don't knock binary & switches. thats where I started. :)

Remember there are 10 types of programmer. Those who understand binary and those who don't.