The Register Home Page

back to article Virtual SG-41 project brings Nazi cipher machine to life in the browser

An enthusiast has built a digital 3D model of the SG-41 cipher machine, replete with wheels, levers, and stepping logic, accessible via a browser. Martin Gillow, the person behind the project, describes the recreation as "part digital preservation, part engineering archaeology, and part 'how on earth did this thing even work …

  1. Aladdin Sane Silver badge
    Pint

    Well done that man --->

  2. virtualcolossus

    More information about the SG-41

    If you want to find out more about this amazing cipher machine, it's history, the technical details and for links to my other simulations (Colossus, Lorenz, Enigma, Bombe, Pilot ACE to name a few), pay a visit to https://sg41.virtualcolossus.co.uk

    1. This post has been deleted by its author

  3. LogicGate Silver badge

    "The use of tools such as Blender to create the 3D models will also help to keep things accessible in years to come."

    Blender, sketchup and other mesh based packages are unsuited for engineering models. The 3d model should be created and stored in a format that supports parametric surfaces (for example nurbs). .step would do so but lose inter-model relations. There may be more modern interchange formats that preserve more information, but the desired format is certainly not .blend.

    1. trevorde Silver badge

      OpenCascade (https://en.wikipedia.org/wiki/Open_Cascade_Technology) might be an option. It's open source and there're a few front end, open source modellers for it too.

    2. Eric 9001

      Don't see anything wrong with .blend - it's good enough for the use case.

      Maybe the FCStd format from FreeCAD would be more accurate?

  4. Ken G Silver badge
    Pint

    Curious

    This is amazing work and I can only repeat the compliment from above.

    It also prompted me to ask, are there any mechanical (or analogue) cypher machines which could be secure in the modern age?

    My limited knowledge of the subject suggests not, since any physical machine can be modelled as a Turing machine and it's process reversed but should anyone know better, please let me know.

    1. Aladdin Sane Silver badge

      Re: Curious

      Obligatory xkcd is obligatory.

      But just because a machine can be modelled and reversed, doesn't mean that it's not secure. Enigma was broken partly because it had design flaws and users were sloppy.

      For true security, it's possible to create automated one time pads.

    2. NoneSuch Silver badge

      Re: Curious

      "It also prompted me to ask, are there any mechanical (or analogue) cypher machines which could be secure in the modern age?"

      Parkinson's Third Law of Computing - Encryption can only delay access to information.

      The medium and the method are irrelevant.

    3. Yet Another Anonymous coward Silver badge

      Re: Curious

      >It also prompted me to ask, are there any mechanical (or analogue) cypher machines which could be secure in the modern age?

      Scrabble bag of letters as one time pad = perfect

      1. Martin an gof Silver badge

        Re: Curious

        Scrabble bag of letters

        W H A T...

        D O Y O...

        U G E T...

        I F Y O

        U M U L

        T I P L

        Y S I X

        B Y N I

        N E

        "Is that it? Forty two? Six by nlne?"

        (From memory, please forgive any misquote)

        M.

        1. Ken G Silver badge
          Joke

          Re: Curious

          keep that up and you'll get hit by a 4by2.

    4. doublelayer Silver badge

      Re: Curious

      In theory, yes, because just as any machine can be modeled in software, any software could be built into a mechanical thing. The EC25519 machine is likely to be rather massive, though, so it's probably best not to.

      If you're asking about actual, historical machines, most of them implemented an algorithm which is not sufficient for security today. It depends where your threshold for sufficient encryption is, because some more complex machines would withstand a basic attack, but if you're faced with people who are motivated to break in, the constraints of historical equipment won't stand up as well as modern software can. Some of that is just the high cost of making keys longer, which in the modern day just slows down the encryption (and for asymmetric keys the generation) stages, but in a mechanical thing means more parts and circuits to handle them. As others have posted, one-time pads are very secure and I'm sure plenty of hardware was built to make the process of using them easier as doing them on paper is annoying and slow, but I doubt that's what you meant.

      1. I could be a dog really Silver badge

        Re: Curious

        A key parameter is what your security requirements are. If it takes (say) a week to crack the message, but the message is about an event happening tomorrow, then even relatively weak encryption may suffice. Many applications only require "real time" security - e.g. long enough to secure a session with your bank and after you log out, it's all done with. But then you have complications regarding what use any session information may be if it is cracked (say) half an hour later, which is very application specific.

      2. Bebu sa Ware Silver badge
        Windows

        Re: Curious

        A physically implemented Turing machine could also implement EC25519 but one would probably need a light year of tape, a pretty robust read/write head and a lot, really a lot of patience.

        At some point the perversity of trying to implement modern systems in decades old tech rivals trying to build a Pentium 4 using 74xxx TTL parts. ;) The reverse of implementing old systems with modern tech is eminently more practical but not without its own challenges.

        Although an implementation of Enigma on a Turing machine would have a certain historical symmetry.

      3. PB90210 Silver badge

        Re: Curious

        One-time pads are fine provided you don't print thousands of identical pads... or reuse pages because of shortages/stupidity

        I've seen both cited as failures by the Ruskies...

        1. doublelayer Silver badge

          Re: Curious

          A good encryption algorithm can still be weakened by using it in the wrong way. That's not what the Russians did there, since the one-time aspect of a one-time pad is very much not optional.

  5. mevets

    Can it decode Trump?

    It would be awesome if it could turn that ooze into something coherent.

    1. Anonymous Coward
      Anonymous Coward

      Re: Can it decode Trump?

      Nooooo

      Don’t poke the ooze!

    2. Bebu sa Ware Silver badge
      Coat

      Re: Can it decode Trump?

      alias trump='cat /dev/urandom | gzip | base64 | rot13 | uuencode'

      # Yes, the garbage output doesn't stop and you cannot precis random nonsense.

      /usr/games/fortune make more sense than anything associated with the current administration.

    3. PB90210 Silver badge

      Re: Can it decode Trump?

      A way to waste time is to find videos of his speeches read by a child or even just a normal person...

      Also check out Michael Spicer's 'The Room Next Door' on YouTube... he plays the guy shouting prompts into Trump's hidden earpiece (explains the terrible hair style) and trying, and failing, to keep him on message and on track

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon