Re: [Add] [Ext] Draft Posting: CNAME Discovery of Local DoH Resolvers

[Paul Vixie <[e-mail]>]

On Wednesday, 1 July 2020 22:09:02 UTC Michael Richardson wrote:
> Paul Vixie <[e-mail]> wrote:
>     > do you see that problem as specific to IoT, or disjoint from RDNS
>     > settings? (i do not.)
> 
> In the IoT case, we need a good, autonomic mechanism to deploy network
> configuration to devices in ways that they really understand.
> (Such as your proxy-HTTP setting, which I entirely approve of).
> 
> Getting the wrong RDNS settings on an IoT device means that it probably does
> not function well (or maybe at all), and it becomes very difficult to do
> RFC8520.  (see my draft-richardson-opsawg-mud-iot-dns-considerations-02 )

speaking as someone who had to create a local responder for 8.8.8.8 in order 
to get my chromecast to believe it had internet connectivity, i think we 
should consider that in the IoT world, there will never be profit from the 
delivery and sale of these devices, only from the services they can offer, and 
the phone-home telemetry they can monetize. google's chromecast already 
ignores the RDNS settings it gets from my DHCP service. therefore, "wrong" is 
in the eye of the beholder, and the interests of IoT buyers are unaligned with 
the interests of IoT sellers. we should consider this negotiation a "hostile" 
one.

> Getting the wrong RDNS settings on a *desktop* browser is a qualitatively
> different thing in my opinion.  It probably operates, but for some
> uses/users it may be a problem, either because of the click-trail it
> leaves, or because of the censorship that the user experiences.
> 
> I think that this is particularly acute for *desktop* browsers, as they do
> not in general have the luxury of just picking a different coffee shop.
> The people who use them are generally vulnerable in a variety of different
> ways.

at the moment, i challenge your specificity here. laptops and smartphones and 
datapads all behave the way you're describing for desktops here.

in the future, i will challenge any specificity on this topic, since all nodes 
whether embedded or operated will have the same capabilities and constraints.

> It is *important* for the browser to be able to clearly indicate what level
> of privacy is being experienced.

i expect that you can get universal agreement on that point. in any case, +1.

-- 
Paul