Re: [DNSOP] [dns-privacy] [Doh] New: draft-bertola-bcp-doh-clients

[Stephen Farrell <[e-mail]>]

Hi,

On 14/03/2019 00:07, Michael Sinatra wrote:
> On 3/13/19 1:43 PM, Stephen Farrell wrote:
>>
>> (dropping dprive list at WG chair request)
>>
>> Hiya,
>>
>> On 13/03/2019 20:29, Brian Dickson wrote:
>>> The starting place for the conversation needs to acknowledge this, and
>>> accommodate it. It is entirely possible that a DoH client that doesn't do a
>>> minimum level of getting user acknowledgement before violating policies,
>>> laws, or contracts, might itself be illegal in some jurisdictions
>>> (jurisdictions that could include some US states, some western countries,
>>> some larger entities like EU, etc.).
>>
>> I almost agreed with you that people need to ack others'
>> priorities. But the above means I can't agree with your
>> mail as "might be illegal" is vastly overstated, there
>> being no relevant difference between DoT and DoH clients
>> in this respect. 
> 
> I believe that the issue of protocol obfuscation that I mentioned
> earlier in the draft-reid-doh-operator thread[1] is a relevant difference.

I do not believe that is relevant to the claim I was disputing
which was essentially that somehow DoH "might be illegal."

If you think your point above is relevant, please consider Tor,
whose major funder is (or historically was) a government and
which is not illegal as far as I know (in many places). And VPNs
too (but only the good ones:-).

> There is another technical issue, and that surrounds the question of who
> is the user and what capabilities does the user have to manage their
> devices.  This has been touched upon with the discussion on opt-in vs.
> default and with Paul's discussion of data exfiltration.
> 
> In my home, I have an "Internet-capable" washing machine.  Of course my
> "smart" TV wants to be on the Internet.  My Foobot *must* be on the
> Internet just so I can monitor the air quality in my own home.  I don't
> want the washer on the Internet at all, and for some of the other
> devices, I want to control what they do on my home network.  With
> embedded and "IoT" devices, there may be limitations on how I--as the
> user--can control them.  There may be hard-coded defaults that are
> difficult to change (and yet have a way of easily resetting themselves
> to "factory default").  Leaving aside for now the issue of licensing
> Ts&Cs, I--as the user--may want to have more *technical* control over
> the devices than their vendor is willing to give me.  One way I can
> assert that control is via the network.  On my home network, I am one of
> the users and I am also the network admin.  I want to assert control
> over the devices for which *I* am the user, but the people who designed
> them didn't give them sufficient knobs for me to do this on the device.
> 
> Another word for software which does things on the network outside of
> the user's control is "malware," whether it is legitimate or not, and I
> realize it predates DoH.  But DoH legitimizes protocol obfuscation at
> the network layer and makes it potentially harder for me to control the
> devices for which I am the user.  So if the goal is to give users more
> control, I'd assert that DoH, at best, works both ways.
Those seem like unrelated (and repetitive) points, except for your
attempt to try equate (I assume) a browser using DoH with malware.
That's the kind of overblown statement that detracts from any other
reasonable points you may make (for me at least).

I do agree that knowledgeable network owners, especially home network
owners, ought be able to exercise control over the networks they own.
I'm perfectly fine to do that in my home network and have no fear of
DoH at all - I'm well used to turning off, working around or living
with what I consider crappy features of browsers (e.g. cookies, JS)
and other tech artefacts. DoH at least has some upsides if it gets
implemented properly.

I don't personally know how to properly and fairly handle such issues
for network owners who quite reasonably don't know anything about the
tech. ISTM that (to date) we've all contributed to failing such network
owners. DoH is nothing special in that respect, nor is RPZ and nor are
many other technologies we've developed.

From my POV, the only thing I hear about DoH that's new(-ish) is a fear
that browsers will turn it on by default in a silly manner, with some
negative but not world-shattering consequences for folks who have a
quite reasonable interest in existing DNS-based technologies and
services.

(Well, that and I've a continuing concern that somehow DoH might end
up enabling web-severs succeed at drive-by attacks on client DNS
caches if someone does some really stupid implementation stuff. But
everyone keeps telling me that'll be ok;-)

Cheers,
S.

> 
> michael
> 
> [1] https://mailarchive.ietf.org/arch/msg/dnsop/Qole4yY0q_-psyrvWabaRAD8_Vc
>