Re: The future of forward proxy servers in an http/2 over TLS world

["Adrien de Croy" <[e-mail]>]

------ Original Message ------
From: "Roland Zink" <[e-mail]>
>
>A proxy is an additional 3rd party which I may trust and want to use 
>(for some service it provides) or I have to use, which is the bad case. 
>In the first case I don't necessarily need the information about all 
>the proxies involved but of cause I want the to know about the latter 
>case.
>
So you're a 6yo at a school, and your school forces you to connect to 
the internet through a proxy so they can block unsavoury/inappropriate 
content.  Please explain why this is "bad"?

I think we need to be careful not to let personal prejudices about what 
constitutes good or bad drive protocol standards.

There's a lot of prejudicial language on this list especially about 
interception, MitM (whether of plaintext http or https).

Sure I can sympathise with a view that a foreign agent covertly 
intercepting and accessing data is "bad", but I can also see legitimate 
uses for interception, such as protection of children, protection of 
intellectual property (DLP) etc, which is "good".

Fundamentally a company that employs people who have internet access 
will decide whether it will block or not.  Many will, and it is their 
unalienable right to control use of their property / resource.

At the moment the move to https everywhere and the abysmal user 
experience of this in most browsers means that there is a lot (and 
increasing amount) of time wasted in organisations dealing with 
misleading/false error messages in browsers.

As a first step, I believe something needs to be done about the error 
messages, and ongoing we need to deal with how to block in a way that 
does not reduce user security, even if we don't move past CONNECT as a 
way to access https.  Whether that requires trust of a proxy or 
something else is another matter.

When I say trust, that comes with consequences, such as believing the 
proxy when it asserts something.  This is a consequence of trusting it.

Adrien

>
>Is it really bad when a browser allows me to trust a proxy?
>
>Regards,
>Roland
>
>
>Am 28.02.2017 um 22:26 schrieb Poul-Henning Kamp:
>>--------
>>In message <[e-mail]>, Kari 
>>Hurtta writes:
>>
>>
>>>Yes, this looks like common sense. And still no one browser does 
>>>that.
>>I think common sense can be summarized as follows:
>>
>>1. If any proxy is involved, both the client and server should know
>>     that, so that they can judge for themselves to what extent they
>>     want to trust the proxy.
>>
>>2. A legitimate proxy has no reason to try hide its own existence
>>     or to deliberately reduce the security, privacy or integrity of
>>     the communication, beyond what is required for doing its job.
>>
>>3. It should be as hard as possible to insert and hide an ilegitimate
>>     proxy (=MITM attack) which undetected can impact security, privacy
>>     or integrity of the communication.
>>
>>On the client side the only politically reasonable and neutral
>>solution is to announce the precense of a proxy by inserting a
>>prominent identification of it above the address bar, so that the
>>user sees:
>>
>>  BIGCORP Inc. Proxy (Contact IT/Bill x1234) inspects this connection
>>
>>  https://mybank.com/
>>
>>Or as it may be:
>>
>>  ELBONIA Government National & Child safety Proxy inspects this 
>>connection
>>
>>  http://bikeshed.org
>>
>>The "proxybar" should be suitably decorated to indicate if the
>>connection to the proxy has any kind of privacy and if there is any
>>reason to think that the proxy really is who it claims to be.
>>
>>The current "political" stance by the user agents means that literally
>>millions of people are behind proxies, legitimate and ilegitimate,
>>without knowing it or being able to see it, without significant
>>X.509-skillz.
>>
>>As for the political fight for a fundamental human right to privacy:
>>
>>More people would probably pay attention, if they could clearly see
>>who tried to mess with their communication.
>>
>
>