[Cryptography] A naming and key distribution infrastructure for the Mesh

Peter Fairbrother peter at tsto.co.uk
Sun Sep 27 23:32:13 EDT 2020


On 22/09/2020 16:55, Phillip Hallam-Baker wrote:
> What is not OK is the SMTP model where 
> the email address is tied to the service provider so that changing email 
> providers incurs an enormous switching cost.

My email used to be **** at zen.co.uk, but I changed it to **** at 
tsto.co.uk.

I did not own zen.co.uk, but I now own tsto.co.uk - or at least 
exclusively rent it from ICANN/Nominet/my registrar or whoever with a 
sort-of guarantee that as long as I pay the rent I can't be evicted.

DNS registry entry service for tsto.co.uk is provided by the people I 
rent tsto.co.uk from.

I can choose anyone I like to provide my actual email service. I don't 
do it myself because running a low latency email service is a serious 
hassle, what with spam blacklists etc., but I could if I wanted to.

> User's can't be in control if they are chained to the 
> service provider.

Which is exactly what is happening in your system, except it is the name 
provider, not the messaging service, to whom the user is tied.


In the m-o-o-t name system there is are no ties at all. Anyone can 
operate a registry and the user can use any registry service. Registries 
which are noticeably incomplete will soon go out of business.

Names are generated at random by the user. He doesn't get to choose the 
name [1]. Random names mean no trademark disputes, and particular names 
are not especially valuable.

If we get a bit ridiculous, we can even have say 8 x 5-bit character 
names. Lots of proof of work needed to prevent forgeries, and then there 
are collisions - except that two people can have the same m-o-o-t name, 
it doesn't matter, as the person's human/user name (generated by lookup 
from the m-o-o-t name) is always displayed when the m-o-o-t name is used.


Capitalist despot, you are still fixated with stickiness and money, and 
your project will not work unless you give it up. !!!  :)

(still a smidgin of truth in that though)


Peter Fairbrother

[1] Except people can do a lot of work and generate a m-o-o-t name with 
a few chosen characters. 8 5-bit characters is probably too small, but I 
think 12 might be OK. 15 or 16 characters are probably OK.


More information about the cryptography mailing list