Discuss mailing list

[Quinn Wood <wood.quinn.s AT gmail.com>]

On Sun, May 26, 2013 at 12:20 AM, Jeff Taylor <shdwdrgn AT sourpuss.net> wrote:
> The way I see it, you wouldn't necessarily need a web-of-trust model,
> although it might be more reliable.
>

> For storing the records, it wouldn't really be that much different that
> what we have now -- using standard zone files -- there would just be a
> lot more of them, or there could be a single large file containing the
> pointers to all the domains.
>

I had posed a question in IRC a while back about mirroring the entire
root zone. I was actually trying to examine the worth of a PGP WoT
like DNS system.

The idea was to use it as a secondary DNS server, using some kind of
local caching nameserver. Instead of just caching though, the rough
goal was to go through a bit of a flow:

< Use WoT Data? >
             |
           Yes
             |
< Record Found? > -- Yes -- ( Request vote: Correct Record/Incorrect
Record, if enabled in user settings) -- Return Record
             |
            No
             |
< Use Legacy Data? > -- No -- Return NXDOMAIN
             |
           Yes
             |
< Record Found? > -- No -- Return NXDOMAIN
             |
           Yes -- ( Request vote: Update Record, if enabled in user
settings) -- Return Record


The primary two problems I faced with the idea (other than simply not
knowing how to approach it technically) where the voting mechanism and
the data storage. A blockchain based system like Namecoin, I feel,
could come extremely close to solving both problems, but it wouldn't
be a web of trust. I thought, and still think, a democratic function
in at least so far as consensus tracking is a precondition to an open
name resolution infrastructure.