« Christchurch Code Camp 2007 - WPF data binding | Main | Usability and the early typewriters »

November 15, 2007

Colossus reconstruction goes live

BBC News:  "For the first time in more than 60 years a Colossus computer will be cracking codes at Bletchley Park."

This is tremendous.  I've recently been reading a couple of books on the Colossus story -- a story long played down in favour of the cracking of Enigma, and which it seems is still only just emerging.  Because of the relative lack of publicity (and even continuing secrecy) around Colossus, I didn't really appreciate the significance of Colossus when I visited Bletchley Park in 2004 and saw the reconstruction in progress, but the machine was awe-inspiring and a bit scary nevertheless.  (Those paper tapes go really fast.)

Rather surprisingly, one of the experts on Colossus is based practically next door in sunny Christchurch.  I wonder if he does speaking engagements?

November 15, 2007 in Software | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Colossus reconstruction goes live:


Didn't someone in NZ invent the transistor (for suitable values of 'invent' and 'transistor') first?
Then of course there is Leslie Comrie - http://en.wikipedia.org/wiki/Leslie_Comrie

You may say that the story of Colossus is played down, but this is as to nothing to the degree many things in the history of computing are overlooked - for some reason cryptography and so forth have a completely disproportionate number of books, journals etc. compared to other stuff (e.g. tabulators, Emmanuel Goldberg).
Most bizarrely, many books on computer history, even ones that are clearly well thought of and sited in other books are out of print - and the only book on Konrad Zuse (in English anyway) is an out of print translation of an autobiography.

The timeline on the BBC page seems typical of their coverage of technology (i.e. pretty crap really).

Posted by: Harvey Pengwyn at Nov 16, 2007 2:03:17 AM

Yes, that is fair comment. But realistically in popular terms there is going to be a lot more interest in hothouse environments like Bletchley Park or Xerox PARC (especially when you have a cast of characters like Alan Turing), and episodes of The Fate Of Nations Hanging By A Thread are always going to be of greater interest than how the Lyons Electronic Office automated the payroll for a firm of tea shops. Unfair, but there you have it.

Posted by: Ivan at Nov 16, 2007 7:33:20 PM

Ah, you raise another interesting question. Given that people had been doing payroll for decades with tabulators, what were they doing with the LEOs that couldn't have been done with them? For my money, given the existence of tabulators, the thing that distinguishes a computer is the conditional branch - one could even effectively do 'joins' with punch cards, there was some sort of device that took two sets of cards and punched values from B onto A based on a lookup - this is covered in Eckerts fine book 'Punched Card Methods in Scientific Computation'.

Stafford Beer claims that he was the first person to use a computer for O.R. problems at United Steel, but were they using them for some sort of allocation problem?

I have read a book about LEOs but my memory fades and I'm not sure it actually was very good on what they did.

Posted by: Harvey Pengwyn at Nov 16, 2007 11:39:51 PM

I was surprised to learn that Colossus didn't have conditional branching, which to me made it less of an early computer and more of a very smart and fast tabulator. Given the constraints and context of Colossus' design, this isn't a flaw: the decisions might have been better left to human operators, and the additional complexity of automating the decisions would almost certainly have made the project nonviable.

The reality is that, as with all developments, there is a continuum between tabulators and computers, and quibbling about whether Colossus, EDSAC, LEO or whatever was the first "true" computer is an occupation best left to blurb writers.

(Also, regarding your question "what were they doing with the LEOs that they couldn't have done with [tabulators]," it may be that they weren't doing anything that was previously impossible, it may be just that the electronic machine enabled them to do it faster and more cheaply. Plus, as you know, LEO did a bunch of stuff: it's possible it was originally designed to do things that weren't practical using tabulators, and then hey, scope creep *grin*. You clearly know a lot more about this than me though!)

Posted by: Ivan at Nov 17, 2007 8:26:09 AM

Well, my understanding is that computers cost a lot more than tabulators at that stage so they might have been faster but certainly not cheaper. However, as you say, once you have one you may as well keep it busy.
Actually, there is a theory (citation needed) that one of the senior guys at Lyons had been at Bletchley Park so knew what could be done, but obviously couldn't say this so went on a trip to America and came back saying 'having seen these American chappies I have had a cunning plan for an idea that I think might have some potential'.
One of the irksomely expensive books may cover this. It seems to be a POD book from OUP... however, this being Oxford you can't order it on-line :-)

Posted by: Harvey Pengwyn at Nov 17, 2007 9:31:34 AM

PS, the first computer... yes, kind of. One could sensibly argue that it is a continuum and there is no first computer, steam engines come at steam engine time and all that, but I think there are lots of things you can't sensibly argue are the first computer because they didn't do anything that tabulators or Goldberg's statistical machine or the thing made by the Norwegian no one has heard of, or indeed analog computers unless you are saying 'digital, did before - unless the argument is that the substrate matters and it has to be relays / valves / badgers to count.

Posted by: Harvey Pengwyn at Nov 17, 2007 11:27:15 AM

Jack Copeland? Speaking engagement?

Are you serious? Because if you get a bunch of people together who want to listen to Jack talk about Bletchley Park and offer to fly him up then my guess would be that he's highly likely to come. As long as it doesn't involve, well, early mornings.

Jack supervised my honours thesis (on Turing's neural networks paper from 1948), he's a marvellous chap. I must get around to buying his book on the ACE sometime.


Posted by: sean broadley at Jun 7, 2008 9:49:34 PM