[Cryptography] A naming and key distribution infrastructure for the Mesh
Peter Fairbrother
peter at tsto.co.uk
Thu Sep 24 20:53:26 EDT 2020
On 24/09/2020 19:09, Phillip Hallam-Baker wrote:
> Underneath, your name is a sequence of 24 random Base32 characters...
> That is the fingerprint of your public key with a work factor of 2^112.
That is too long for people.
For m-o-o-t [1] I devised a 16 random base32 character identifier and
address and (sort-of) telephone number.
Chap makes a file with his name, digital signature, email address,
telephone number, snail-mail address, whatever he wants to include in
it. He makes a hash of the file, then does a 2^20 proof-of-work a la
bitcoin and truncates the result to 80 bits.
That is 2^100 bits work factor, and it is very hard to find a preimage
for a given hash, needed to do most MITM attacks. You can find
collisions with a lot of effort, but there isn't much point - eg the
name will be wrong.
The file is published in a directory, indexed to the identifier. Anybody
can publish directories. You can even publish your own directory.
If someone wants to call you, or email you, or send you a parcel, you
give him the 12-character string. Eg m-ahr7-dt46-j37f-hgsb. [2]
Sufficiently memorable, capable of being spoken or written as well as
QR'ed or barcoded.
Done properly, when sending someone an email, or printing an address
label, or calling someone on the 'phone, the address translation from
identifier to email address, snailmail address, telephone number etc
would be done automatically in software.
The software would also display the name and check the PoW and hash, and
eg for telephone calls and email would use the key to encrypt.
I think it might even be possible to go to 12 base32 character
identifiers if you increased the PoW to 2^40 - you could pay a bitcoin
miner a dollar to do that for you if you can't do it yourself.
Key distribution is by identifier.
> Thats an OK way to label things internally. But it really isn't
> something most users will find usable. And thats fine. This can be a
> naming infrastructure that is only used by the 99% not the technorati.
Why should the technorati use a different system? They will usually be
talking to the 99%, except maybe helldeskers and bofh..
Peter Fairbrother
[2] they all started with m- so you would know what they are, like all
telephone numbers in the UK start with 0, and internet addresses have a //:
[1] m-o-o-t was/is? a communications ecosystem which only allows secure
communications, originally based on a bootable CD. I am not a good
enough coder to get it to work securely, reliably and user-easily, which
is why it is a bit moribund, though I believe the design was sound.
Perhaps one of three.
More information about the cryptography
mailing list