bye bye ViaCode

[Brian Gladman]

From: "David Hansen" <davidh@spidacom.co.uk>
To: <ukcrypto@chiark.greenend.org.uk>
Sent: Monday, June 10, 2002 10:29 AM
Subject: Re: bye bye ViaCode


[snip]
> Only if you trust the centralised system. Even then you don't know,
> you only hope, though the eminently sensible web of trust can have
> similar problems.
>
> Web of trust works the way humans work. It is possible for this to go
> wrong, but it is a darned sight better than the sort of inhuman top
> down system  that some "experts" seek to force upon people.

Well said David.

When applied to provide secure information exchange between people, it is
obvious that WoT can only be effective if its use is limited to the
provision of automated support for pre-existing human trust relationships.
When limited in this way WoT can work well but this means that the webs
involved are likely to be domain specific and limited in terms of their
overall size.

The fact that WoT is often advocated (or even used) for more than this is a
failure of those involved in this advocacy and not a failure of the model
itself.

For e-commerce support the existing CA model was bound to fail because it
placed responsibilities in the wrong place at the wrong time.

But far better models are now emerging in which the relying party approaches
a certifying party *** at the time of a propsective key dependent
transaction *** to detemine if the key(s) in question is/are valid.  This
provides a full context in which key validation can take place and hence
removes many of the hurdles to effective key certification.

Whereas a CA can hardly underwrite a key in the face of all unknown future
transactions, a CA may well be prepared to underwrite a key for a specific
imminent transaction since the risks can be evaluated and largely contained.

In my view these emerging CA models will be far more successful than those
being pushed at the moment.

  Brian