Re: [Add] fixing coffee shop brokenness with DoH

[Eric Rescorla <[e-mail]>]

I think we're now entering territory that we've gone over before and that
is largely off topic for this list. so I'm going to opt not to revisit it.

-Ekr


On Thu, Jul 25, 2019 at 8:00 AM Paul Wouters <[e-mail]> wrote:

> On Thu, 25 Jul 2019, Eric Rescorla wrote:
>
> > For context, our breakage budget here is very small: any change to
> clients which causes connection error
> > rates to increase by more than fractions of a percent is not really
> something we can deploy.
>
> What percentage of hijacked connections due to maliciously modified DNS
> is within your breakage budget? In other words, how do you balance security
> versus user convenience? A certificate that has expired only two hours
> ago seem to be fine to hard fail on. But a broken DNS transport is not?
>
> There is a bit of interaction here. People messing (or neglecting)
> their DNS of endusers might do this less if they get flagged in common
> browsers for doing so. Of course I symphatize with you that you do not
> want that to become your problem. But how long are we kicking this
> can down the road?
>
> Everything on the internet should be authenticated. We keep delaying
> this as a hard requirement for DNS only, for increasingly less convincing
> reasons.
>
> Paul
>
> --
> Add mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/add
>