Re: Proposal: Marking HTTP As Non-Secure

On 2/8/2016 8:37 PM, Eric Mill wrote:
>
>
> On Mon, Feb 8, 2016 at 10:41 PM, Kyle Hamilton <[email protected]
> <mailto:[email protected]>> wrote:
>
>
>
>
>     The thing that stopped me was "mandatory payments to CAs".  This was
>     another decision by "people who knew better" attempting to enforce a
>     security policy on those for whom the one-size-fits-all security
>     policy
>     didn't fit the needs of.  
>
>
> I understand this resentment, but I think given widespread inexpensive
> DV (even before Let's Encrypt), it's not much different than paying a
> nominal fee to a ICANN-accredited registrar for a domain itself. The
> $10-12 fees for domains and certs really are pretty nominal, even for
> folks with no income or lots of debt (which describes me for most of
> my life). It's a lot different than the $99/year fee you pay for the
> chance for Apple to approve your app.

That certainly applies to the current situation, but I was responding to
the "all traffic should have been encrypted from the beginning"
sentiment.  Back in the 1990s, with Netscape beta 0.94 (which did not
have support for adding CAs, and which tried to position the "unbroken
blue key" as "you only need this if you're talking to your bank"), it
was as onerous as EV requirements now seem to be.

> Those registrars aren't like, morally superior. They have a deal with
> ICANN, who holds them to non-legally-binding but technically
> enforceable standards (just like browsers do with CAs), and I'd bet
> lots of money their operational practices are just as shady underneath
> the hood as are many CAs. The internet's a messy bunch of messy
> companies and it does cost money, but we still have a system where you
> don't have to pay much to play as a domain owner for the basics.
>
> And if you don't need "public trust", because *both* parties are
> informed and able to rely on non-publicly-trusted certificates to
> communicate in a way they wish, that's always still possible, and
> nothing Mozilla or Chrome is doing affects that. It's when only one of
> those parties (the server) is making an informed choice that browsers
> have a strong opinion.

Nowadays, certainly... but Mozilla still likes to do the "scare the
lowest common denominator" language of "Legitimate sites will not ask
you to do this!" in the add-an-exception dialog.

>     >> Would you write everything on a postcard? or would you at least put
>     >> the letter in an envelope?
>     > People choose either/both all the time.
>
>     Yes, but do you ever see a bank sending a statement on a postcard?
>
>
> I'll add that one major difference here is that ordinary people have
> the capacity to evaluate the privacy tradeoffs in whether they choose
> a physical envelope or postcard. They understand the risks a postcard
> carries, and they can see the metadata that's carried on an envelope
> either way.

More importantly, it's the sender who does this, and the knowledge that
the content was openly available on a postcard is instantly
comprehensible to the recipient.

> People have no concept of the tradeoffs or metadata involved in online
> communication, which creates a truly corrosive lack of transparency
> and accountability. Users are treated by various governments and
> corporations who run invisible services however those governments and
> corporations wish to treat them, and they can rely on legislative and
> regulatory bodies to have a very small % of people on staff who
> personally and completely understand the issues at play.
>
> It is fair to describe the use of a physical postcard as an informed
> choice. No one (of statistical significance) makes an informed choice
> to use plaintext HTTP as they go about their day.

This is correct.  But, again, the use of plaintext HTTP is even more
disturbing given that there's no physically examinable substrate. 
Unencrypted HTTP can be modified and forged in flight, and there
wouldn't even be the challenge of causing it to look the same to closer
examination (since there's no implementation that reports on the number
or size or jitter of TCP packets, and HTML can be arbitrarily padded
with whitespace with no difference in the presentation).

Unencrypted HTTP should be explicitly marked in the user interface as
worse than insecure, since we'd be marking aNULL-eNULL as insecure under
the same rules (and even then, that at least provides a modicum of
integrity protection on the wire, with some digest being used for the
MAC.)  Arguments that "it's never been marked that way before" only
underscore the fact that people don't actually understand what's really
being made available to prying eyes and malicous man-in-the-middle actors.

>
> -- Eric
>
>
>     _______________________________________________
>     dev-security mailing list
>     [email protected] <mailto:[email protected]>
>     https://lists.mozilla.org/listinfo/dev-security
>
>
>
>
> -- 
> konklone.com <https://konklone.com> | @konklone
> <https://twitter.com/konklone>