W3C home > Mailing lists > Public > public-privacy@w3.org > January to March 2015

RE: On the european response to Snowden

From: Mike O'Neill <michael.oneill@baycloud.com>
Date: Wed, 28 Jan 2015 09:12:23 -0000
To: "'David Singer'" <singer@apple.com>
Cc: "'Danny Weitzner'" <djweitzner@csail.mit.edu>, "'Rigo Wenning'" <rigo@w3.org>, <public-privacy@w3.org>
Message-ID: <3a1401d03ada$8dad49d0$a907dd70$@baycloud.com>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David, comments to your comments inline

> -----Original Message-----
> From: David Singer [mailto:singer@apple.com]
> Sent: 27 January 2015 14:33
> To: Mike O'Neill
> Cc: Danny Weitzner; Rigo Wenning; public-privacy@w3.org
> Subject: Re: On the european response to Snowden
> 
> Thanks Mike, comments inline
> 
> > 1) Signalling.
> > 	We saw a bit of this in the DNT discussions. How to create a signal
> conveying a user's explicit agreement for something or their preferences for
> something to one or more entities that may exist across multiple origins, in a
> secure untamperable way. This may eventually be superseded by:
> 
> A challenging problem.  These signals and preferences tend to be small, and
> padding them and then signing them digitally would seem to be using a
> sledgehammer to crack a walnut.  But maybe the walnut is growing in
> importance.  Other ideas?

I was meaning more the general problem of signalling between entities, i.e. between the UA acting for an individual and companies which control many domains/origins. There are several use-cases that came up in DNT and it requires authentication of identity which was also why it will be subsumed into point 2. 

> 
> > 2) Anonymity.
> > 	To ensure privacy we should be able to trawl the net anonymously, but
> with some identity available through defined transactional processes. For
> example we may allow a subset of our identity to be discovered by some parties
> we know about and have reached agreement with. This might just be a broad
> audience categorisation (male, geek, whatever) or it might be more specific
> (MEP, a particular child's parent, member of a club). Visible identity changes with
> circumstances i.e. I could anonymously apply for a loan or agree to pay for a
> purchase but I would need to be accountable. My legal identity would have to be
> discoverable in certain agreed circumstances. We may also agree, through
> membership of a "rule of law" jurisdiction ,that our identity is discoverable by
> law enforcement under agreed (by society) circumstances.
> >
> > This may go beyond HTTP, i.e. IPv6 anon. auto configuration everywhere or a
> new internetworking layer, focus on stopping fingerprinting, and it is a big one.
> It will need heavy guns.
> 
> Online anonymity — secrecy — is hard, as you know. ToR is hardly an easy or
> universal solution. I recently did the thought experiment “what if every router
> was a NAT box?” — this would mean that IP addresses would be useless as
> proxies for identity — and the answer is that anonymity would improve but
> many other things (e.g. phone calls) would suffer. Again, ideas for this would be
> good.

I think there should be an out-of-band identity exchange, non-trackable i.e. does not use UUIDs but established below the tunnel. Maybe in the https handshake or in an internetwork layer.
The identity exchange should be under the control of both parties, but also visible to third-parties in defined circumstances for instance when accountability or law enforcement is required.

> 
> > 3) Encryption.
> >
> > There is talk about making end-to-end encryption illegal. While this may seem
> silly and is probably a shot across the bows, https everywhere stirs the hornet's
> nest. I think an answer involves some process whereby https is made more
> secure (via certificate pinning etc.), available to anyone but that law
> enforcement is given the means to determine identity through an internationally
> agreed process i.e. along the lines of 2).
> >
> > I think any backdooring process will just end up helping the bad guys, so we
> have full ETO encryption available but if the net can properly ensure privacy and
> security only a minority will need it.
> 
> So you envisage encryption that is end-to-end and backdoor free, but
> nonetheless accessible to lawful intercept. Challenging in today’s environment,
> but maybe there is a solution.

I was thinking more that the identity was visible to lawful intercept, not necessarily the encrypted content. But if privacy and security are guaranteed without encryption then there would be less need for it. I forgot to mention integrity, there should be a way to ensure integrity of the data (such as javascript) transmitted between mutually identified parties, without having to put everything through an encrypted tunnel.

> 
> David Singer
> Manager, Software Standards, Apple Inc.
> 

Mike
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.13 (MingW32)
Comment: Using gpg4o v3.4.19.5391 - http://www.gpg4o.com/
Charset: utf-8

iQEcBAEBAgAGBQJUyKf2AAoJEHMxUy4uXm2JlVcH/1r0l9iUX8wPdcBN2kgM/Eim
u3Nzs3K2I72jSMju5Mih2Cpt43/qr/B3LxNSPnE+ZZezHfTMqTF9xGgL1GK1zN4b
i0FabfQ1yDVANpH3pPnUqSgpUvcxtbg2bXOjsJkwTRkjCg+hOmqvz0DoLh+8MARM
rdkpwYV8SiLlcIM5gKwTfm6TrDIs1T2Xl1D7qUIaXJ2JF98SeHoaizGcgCgUN1Kd
RhMEaXM+Wa3gQ/ZtrJgea9YkyNiBkBK9xuHdpcq4aTiCVmY8ZrCtzjv5F0ddWTWa
Z1hxD9svQ6Z9bNhZRKe2r93F63pzfnEmY3buNL6DZrk4IdtQwHsjveKbr9I90qo=
=3sOn
-----END PGP SIGNATURE-----
Received on Wednesday, 28 January 2015 09:14:19 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 28 January 2015 09:14:20 UTC