- From: Ryan Sleevi <ryan-w3-web-security@sleevi.com>
- Date: Tue, 1 Jul 2014 10:08:06 -0700
- To: "Rigo Wenning" <rigo@w3.org>
- Cc: public-web-security@w3.org, "GALINDO Virginie" <virginie.galindo@gemalto.com>
On Sun, June 29, 2014 4:44 am, Rigo Wenning wrote: > On Saturday 28 June 2014 05:36:24 GALINDO Virginie wrote: > > Granting permissions to unauthenticated origins is, in the presence of > > a network attacker, equivalent to granting the permissions to any > > origin. The state of the internet is such that we must indeed assume > > that a network attacker is present. > > The error here is that we assume the service/origin to be trustworthy > and the attacker to be malicious. But in case of tracking, the > authentication actually harms. So having more authentication isn't > providing more security for the end user in general. In tracking, the > service you're interacting with is the attacker. How does your model > cope with this and how is it avoiding to switch from tracking to > authenticated tracking? > > Now if we want to talk about origins and trustworthiness of code, how > does your work relate to the Trusted Computing platform? Is it just > basing itself on TLS or is it going further? Or is it just a list of > partial URI-strings that will trigger better permissions? Have you > thought about integrating provenance into the model? > > --Rigo > Rigo, Virginie's forward dropped all of the lists that this was sent to, including the place it's actually being discussed (blink-dev@chromium.org). If you wish for a reply, your best option is to actually send it to a list where it is being discussed. I fear you've misunderstood the proposal, or confused it with something else. With the exception of client certificates (which provide TLS mutual authentication), TLS only authenticates a server to a client, not a client to a server. This notion of "authenticated tracking" is thus a fabrication, because it does not exist, any more than it does for HTTP. Note also that the discussion of "secure origins/transports" is not exclusive towards HTTPS, and includes other forms of code authentication, such as signed extensions. This has nothing to do with Trusted Computing Platform. Again, I suspect there is some confusion about what's being proposed, and you'd be best off seeking clarification where it's being discussed if you feel it's worthwhile. Cheers, Ryan
Received on Wednesday, 2 July 2014 12:27:33 UTC