The Long-Term Future of the Cryptography Policy Debate

[Brown, R Ken]

> Roger Fleming[SMTP:roger@police.tas.gov.au]   wrote, in reply to Stefek
> Zaba:
> 
>  Of course it
> is possible someone will devise a much more efficient device than this "Mk
I"
> quantum computer. The consequences of a processor that is able to
> instantaneously perform massively parallel computations of any size are
> almost unimaginable, and will go far beyond obliterating nearly all forms
of
> crypto.

I really don't understand why it should obliterate crypto. 

Surely, however fast the computation devices are, if the algorithm needed to
decrypt is harder to run than the one needed to encrypt,  decryption can
always be made to take an arbitrarily long time by choosing a large enough
key? Unless you are suggesting that computation will be *so* fast there
isn't enough  storage space for a large enough key, which I find hard to
believe. Even if we were forced to use megabyte keys we could still encrypt
useful messages on CDs or whatever replaces them and transport them by
bike...  ( there is no sign that  network speeds will ever catch up with the
amount of data in storage - the capacity of magnetic media has been
increasing faster than bandwidth for ever (as far as electronic computers
are concerned) and still is. So the ratio between how much data you store
and how much you can send down a wire in a second will carry on growing. So
there will still be a place for the bicycle messenger in the Brave New
World)

Of course it is possible that  quantum devices might be too expensive for
ordinary users to buy them.  If  governments,  armies, banks, the mafia and
so on will be able to afford them they might  be able to decrypt  the
messages that the rest of us send - which is in practice the situation we
have been living in since WW2 (we now have techniques that the rest of us
can use that the spooks can't crack but most people don't yet use them).

If they are cheap enough then we are back where we are now.

My gut feeling is that this won't happen and that the  exponential increase
in the power of a single computing device that we have been used to will
come to an end sometime between 5 and 50 years from now.   When that happens
everything changes. At the moment  big business uses machines only 2 or 3
times more powerful than kids use to play games - and both use machines more
powerful than the military put in the field (it is as if a Scalextrix toy
car had half the speed of a Formula One racing machine and almost as much
room for goods as a 20 ton lorry...) When the increase in chip power is hit
singe computing devices  will be between 10 and 10 million times faster than
they are now (that's a safe bet I think!) and people who want faster
machines will either have to run cold or run clever  (by getting into
massively paralel computers). In either case we are back to big machines
with real plumbing - I suspect that  the supercomputer (if not the
mainframe)   will have as large a niche in the 2020s as it did in the 1970s.
But I doubt if they will be fast enough to crack 10 Kb keys in real time.