[Cryptography] [FORGED] Re: Escrowing keys

[Peter Gutmann]

Phillip Hallam-Baker <phill at hallambaker.com> writes:

>My two daily driver cars are a couple of Jaguar convertibles built in 1999.
>They are surprisingly reliable

>My two daily driver cars are a couple of Jaguar convertibles built in 1999.
>They are surprisingly reliable

That's not necessarily a ringing endorsement, "for Jaguars (something where
Lucas Electric was involved), they are surprisingly reliable".  That's like
"this Windows server has an uptime of nearly two weeks" or "I actually got a
second day of battery life for my phone the other day".  I've got a bunch of
PLCs here that date back as far as the 1980s [0], and they're unsurprisingly
reliable.  They were designed to be that long-lived, and the only reason I
don't have older ones is that before that they tended to be hardwired.  So
getting back to the OP, there's hardware out there that not only can be
trusted beyond ten years, it'd be regarded as defective if it didn't last ten
years.  Or twenty.  Or thirty.

Or, in the case of relay ladder logic controllers, eighty or a hundred.

This sort of life cycle is more or less impossible for crypto people to
understand [1].  Conversely, SCADA/industrial control people understand the
life cycle but not crypto.  This is why we have so much SCADA gear that's an
OWASP top-ten antipattern.

Peter.

[0] Using a somewhat loose definition of PLCs, the "embeded computer running
    control logic software" started more in the 90s.
[1] There are a few practitioners that get it, but they're so busy building
    reliable systems that they don't have much time to talk about them.