[cap-talk] crypto purism vs. Pet Names (was: Firefox breaks the principle of identifiability)

[Sandro Magi]

Zooko O'Whielacronx wrote:

> Sandro Magi wrote:
>
>> I don't see why a service like Google is impossible in this scenario. 
>> You
>> sign up for an internet account and your introductory page is your ISP's
>> start page; this is your portal to the internet. Your ISP has already 
>> gone
>> through the hassle of typing in Google's unique key (if this step is
>> indeed necessary) and establishing a YURL to Google. Why not then just
>> search for "Coca Cola"?
>
>
> Indeed, and why couldn't I configure my Firefox that if I type in a 
> name that isn't in my local Pet Name database, it will resolve it by 
> querying verisign.com?


Almost, but a name potentially has a one to many association, not one to 
one like DNS.

> So Pet Names seem like they could be a superset of DNS-style names.  
> This makes me suspicious of them, because they would then suffer an 
> externality in practice which would tend to make for "name 
> monopolies".  This would be the same bad result as we currently have 
> with DNS and Verisign, even though it would arise from the emergent 
> decisions of millions of agents in the presence of an externality 
> rather than being a design decision imposed by the technology itself. 


It's possible, but I think it's more resistant than DNS to such an 
outcome. The name associations are context-sensitive, so anyone with 
better search technology is a potential competitor. This could yield 
indexes that are better at finding businesses or individuals rather than 
general information for instance. This approach seems far more amenable 
to open competition and the emergence of niche markets than enforced 
global domain names. At the very least, it seems an improvement over the 
current system to me.

Sandro