Re: [pkng] some thoughts

[Peter Saint-Andre <[e-mail]>]

On 11/14/09 2:56 PM, Leif Johansson wrote:
> Peter Saint-Andre wrote:
>> On 11/12/09 1:27 PM, Leif Johansson wrote:
> 
>>> 1. bottoms up vs top down
>>>
>>> In PKIX the key relationship is between the key and the CA. I
>>> believe PKNG should turn this around and be designed based on the
>>> relationship between the key and the key consumer.
>>>
>>> I'm consciously trying to avoid using terms like user-centric or
>>> user-managed since they evoke images of specific technologies or
>>> patterns and in fact I'm thinking about the RPKI and the KARP
>>> when writing this.
>> Leif, I skimmed your email while I was sitting in the DTNRG session this
>> morning, so the mention of bottom up rang a bell. PKI is impossible when
>> there is no "I" to be had (distressed networks, ad-hoc meshes, etc.). As
>> people carry more and smaller personal devices, they might want to
>> exchange information directly in their interactions (beamed, radioed,
>> etc.). They won't necessarily have access to canonical sources of trust
>> located in hierarchies or roots. Further, they might not *care* about
>> hierarchical trust, instead putting a higher value on bottom up or peer
>> to peer trust. Perhaps trust will be based more on social networking and
>> personal interconnections (think IM buddy lists, LinkedIn, Facebook,
>> Twitter). Keys and certs (or whatever we end up calling them) are cheap,
>> so people might generate new ones when needed (or for different
>> purposes), check credentials with people or entities in their network
>> only when they interact with a person, etc. We have ways to push this
>> information out (hey everyone, I've got a new keything) and for
>> performing on-demand checks (hey Leif, Paul says his keything has a
>> print of foo, is that consistent with what you know?).
> 
>> Just a few random ideas generated on the Nozomi train out of Hiroshima... :)
> 
>> Peter
> 
>> --
>> Peter Saint-Andre
>> https://stpeter.im/
> 
> 
> Yeah this is how I think about it too - to buzz a word: crowdsourcing
> identity (but we'll have to be careful not to build another pgp that
> doesn't have a business model).

Despite my libertarian rhetoric, I'm not absolutely opposed to (some)
hierarchies. Perhaps what I'd like to see is an approach that is not
exclusively hierarchical (PKI with the focus on "I") or exclusively
peer-to-peer (PGP), but instead an approach in which there can be many
input sources. Some of those sources might be privileged or weighted
more highly by some participants. And perhaps this is not all that
different from PKI (I don't trust all CAs) or even PGP (I don't care if
some random person has signed your key, only people I know about or have
some trust in). Unfortunately I think that X.509/PKI and PGP/WoTs have
been presented and deployed as two diametrically opposing approaches.
I'd like to see us reach some harmony between the two concepts...

Peter

-- 
Peter Saint-Andre
https://stpeter.im/