FW: [Dundi] DUNDi popularity

[Joseph Ashwood]

----- Original Message ----- 
From: "Michael Richardson" <mcr at sandelman.ottawa.on.ca>
Subject: Re: FW: [Dundi] DUNDi popularity


>  You need the PGP model instead.
>  Every DUNDi node needs to be a CA/signer. You can have web-of-trust.
>
>  You may also want to look at SPKI:
>  (Rivest= R of RSA, Ylonen=SSH)
>
> 2692 SPKI Requirements. C. Ellison. September 1999. (Format: TXT=29569
>     bytes) (Status: EXPERIMENTAL)
>
> 2693 SPKI Certificate Theory. C. Ellison, B. Frantz, B. Lampson, R.
>     Rivest, B. Thomas, T. Ylonen. September 1999. (Format: TXT=96699
>     bytes) (Status: EXPERIMENTAL)

That's more or less exactly what I was thinking in addressing Anders' issue, 
then through the certs we have a handoff of trust, and while a single 
failure would simply identify the end node (identical to the IP address 
failure), a large scale failure (the real problem) would allow tracking of 
the bad connectivity point allowing both routing around it, and any 
additional remedies allowed under law.



Incidentally, through a small extension of the certificate (already enabled 
through most of the options) we can include link tranversal costs, to 
address the suggestion made by Jeffrey in a different subthread. With 
increased connectivity these values would tend towards 0, but as an uptake 
measure it might be of value.

                            Joe