• Two observations about the reality of public key cryptography and digital signatures

    by  • June 7, 2008 • The Global Picture • 0 Comments

    1> Unaided, a digital signature only proves that you uttered a statement, not that you originated it.

    2> The mapping of keys to individuals, and individuals to keys, is many to many.

    To get any more than this out of cryptography, you need to start adding things.

    1> A timestamping service which records the first time an artifact was seen allows people to claim origination, but only if there is a universal practice of getting things stamped as soon as they are authored.

    2> A universal biometrics database coupled with key issuance can make the mapping of keys to people one to one.

    The reason that so many cryptographic systems experience profound difficulties behaving in a way that makes them useful to people is that the properties that people wish they had, they don’t have. Proving authorship would be easy… if only… proving identity would be easy… if only…

    A context-free, pure-digital system has extreme limits on what it can achieve in the World of People. Things like sockpuppets are fundamental to the digital medium because of the many-to-many property. Man-in-the-middle attacks are a logical consequence of a digital signature only proving that you said something, not that you originated it.

    However, I think that taken as-is, if you do not attempt to make digital systems do what they do not, you can still get a long way. For example, a digital signature on a piece of text could be thought of as a vote, and with a little additional work, a plausible ballot could come out of it although, again, the question is “who may vote, and how often?”

    You cannot hope to arrange the affairs of the real world with computers. Furthermore, the “natural law” of the digital world corresponds to the natural law of the natural world at almost no point: information can be indefinitely replicated at minimal cost, identity shimmers like rainbows reflected on a lake, and all relationships with time and space are maintained only by diligent effort, lest clocks stray, or IP addresses become mobile or NATted.

    The core problem of computing is that we have been trying to model the non-digital world in the digital world from the beginning, and what we have is a terrain made of horseless carriages and ornithropters. This goes right the way down into concepts like files. One of Unix’s most successful aspects is the pipe, which is a lot less physical than most notions about files and folders. Tags are the beginning of breaking away from the physical in semantic space. Tearing down the attempt to model the physical in the non-physical at the level of intuitive-but-wrong imitation is going to be the first step in doing real engineering on the systems we have built.

    It’s a spinning metal disk with variegated magnetic fields. It’s a bitwise addressable memory. It’s a processor. All this work to fake paper and reflective surfaces on computers? Those billions of lines of code?

    Stop. Just stop. Start at the beginning: input and output. Information retrieval, display, processing, storage and transmission. Simple primitives. Do what cryptography can do, and stop striving to make it do what it cannot.

    This is what I’m saying: http://www.dogme95.dk/the_vow/vow.html.

    We need to throw away the physical models that underlie computing implicitly in every area, from windowing systems, fake paper, fake pages, fake files, fake folders and a million other fake objects, and get down to the basic tasks in as simple and direct a way as possible. Three-wheeled cars, two winged planes, and raw data processing with interfaces which do the job right rather than trying to mimic tools that most people have never used.

    You know that ruler at the top of a Word file, where you can set margins by dragging little pointers? That used to be a piece of metal on a typewriter.

    I used them.

    Nothing on a computer should try to emulate the real world.

    Nothing on a computer should dress up its function to appear like an object from any date before 1950.

    That’s not to say there’s no place for interface metaphors, but not paper. 85-dimensional vector spaces, races of alien intelligences which organize our messages, whatever it takes. But let the virtual go free at last.

    We’ve had all the cushion generations we needed. The kids growing up on Wii are virtual. Give them the tools they need, not jumped up electrical typewriters.

    flattr this!


    Vinay Gupta is a consultant on disaster relief and risk management.


    Leave a Reply

    Your email address will not be published. Required fields are marked *