Re: [saag] Discovery: can it be solved

[Martin Thomson <[e-mail]>]

I'm personally not interested in more novel techniques for getting bits from point A to point B.  PIR has applications - it might be that Tor is one - but I don't think that framing this as sampling from a larger dataset is the right way to approach the problem. Maybe global and globally consistent knowledge is part of an answer, but if topological information were sufficient, then I suspect we'd not be discussing this.  For example, DHCP/RA provides information that is very topologically relevant.

The questions here are more fundamental.  What security basis do we have for clients retrieving critical information from the network? How does that affect the security of those clients? How might that shape how the services that are discovered can be deployed? What does that do to the diversity of those services?

At some level, the question of IP intermediation (aka routing) has been solved.  We assume virtually nothing from IP forwarding and routing, but also expect very little from those providing that service.  (The same is largely true for Tor, with some interesting caveats.)  The services Watson lists provide more advanced capabilities and come with associated risks. For example, DNS resolution creates a privacy exposure for clients. That means that it isn't so easy to allow arbitrary others to provide these services.

On Mon, Nov 15, 2021, at 23:48, Antoine FRESSANCOURT wrote:
> Hello Watson,
>
> I agree with you that discovery is a topic to address in the 
> development and deployment of privacy-preserving solutions. Indeed, if 
> we want to avoid having to retrieve a full view of a network's topology 
> to discover intermediaries / trusted peers, we need a way to make a 
> more targeted discovery secure against deanonymization attacks.
>
> I think I mentioned this during the Q&A of the talk I gave during the 
> PEARG meeting at the last IETF, but some people in the Tor community 
> have presented a way to retrieve information about intermediaries 
> without having to retrieve the whole topology. This work called PIR-Tor 
> can be read here:
> https://www.usenix.org/legacy/events/sec11/tech/full_papers/Mittal.pdf
>
> I think private information retrieval is an interesting potential 
> solution to the private discovery problem, in particular in its 
> information-theoretic form. All in all, I am really interested in 
> investigating this question, and I think PEARG is a good working group 
> to make progress on the matter.
>
> Best regards,
>
> Antoine Fressancourt
>
>
> -----Original Message-----
> From: saag [mailto:[e-mail]] On Behalf Of Watson Ladd
> Sent: Saturday, November 13, 2021 4:57 PM
> To: IETF SAAG <[e-mail]>
> Subject: [saag] Discovery: can it be solved
>
> Dear Saag,
>
> It seems that a large number of proposals, from OHAI, DOH, Privacy Pass 
> and most recently some IP privacy work have received the same kind of 
> pushback around 'discovery': the argument that we need a mechanism for 
> clients to find a large list of available and willing intermediaries. 
> The challenge here is that we have not seen a proposal, and it doesn't 
> seem to answer the real issues about diversity of intermediates, user 
> trust in what they discover, or address any of the very real issues 
> around user experience.
>
> I think this is a necessary discussion to have once rather than 
> constantly relitigate.
>
> Sincerely,
> Watson Ladd
>
> --
> Astra mortemque praestare gradatim
>
> _______________________________________________
> saag mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/saag
>
> _______________________________________________
> saag mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/saag