On 3/19/18 1:08 PM, Selena Deckelmann wrote:
There's a lot of thinking that went into the agreement we have with
Cloudflare to enable this experiment in a way that respects user privacy.
I would like us to be very clear that there are two separate things here:
1) Is this behavior good for users?
2) Will people think this behavior is good for users and for them?
(Maybe this should itself be two separate things.)
Here's how I see this:
* There are some concerns being raised about item 1 (e.g. it may be good
for users in the US but less good for users in jurisdictions where the
legal obligations of ISPs are qutie different). Have we considered
doing the experiment only in some geographies, ones where we can make a
particularly strong case for the status quo being user-hostile?
* For item 2, fundamentally, we want to avoid people feeling like they
are being betrayed when they discover they are part of this experiment.
To me that seems like it requires clear messaging that they _are_ part
of the experiment. If we tell a nightly user "you are part of a DNS
experiment, here are the details, here is how you opt out", that leaves
a _very_ different impression from (hypothetically; I haven't checked
whether this would be a failure mode of the proposed setup) the nightly
user being unable to access some intranet site that they set up
/etc/hosts entries for, spending a bunch of time figuring out why, and
then discovering that we silently changed how their browser does DNS.
In the latter situation people will be predisposed to believe the worst
and not listen to any explanations.
* Assuming we go forward with this, we should very seriously think about
the messaging, both in-product and out-of-product. For example, I would
think that we would want this to appear on tech news sites _before_ we
start doing the experiment, not after. That gives us a chance to
present our case in a non-crisis atmosphere, gives people a heads-up
about what they should expect, and is a lot less likely to be perceived
as us trying to sneak things in.
* A lot of this is about trust; both building trust and destroying it.
Fundamentally, for most people (I'd guess nearly all) trust is not a
logical decision; it's based on gut reactions. Trying to logically
convince people that they should trust us is just not going to work if
their instincts are screaming at them not to trust us. That means that
even if we're 100% sure something is better for users and even if we
have super-convincing arguments for it, we need to seriously think about
the way it's messaged (or not) and the resulting impact on trust. It
doesn't help that what comes across as reassuring to one person comes
across as weaselly information-free double-speak to another....
I'd like the team to share this in a blog post
about the experiment
This seems like a good start, and we may want to then make sure whatever
information we are trying to put out there actually gets picked up by
widely-enough-read news bits so people aren't blindsided.
-Boris
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
