[ux] Proposal: Shared language to empower our efforts

Sat Mar 27 17:18:58 UTC 2021

Hello list, in this email I want to share a way of thinking that can help
us discuss and define a roadmap in Strategy, Design, Tech and Research.

First of all, I think it is difficult to have discussions around the impact
we have in users' lives. But moreso, the ethical decisions we make each
time: (i) we release a version of TB (ii) or create a roadmap for the next
release (iii) or plan a new feature to build, as there are many ways we
could materialize each feature, (iv) and most importantly how should we
research if our intentions are met in real life scenarios.

Having said that, I believe it would be* extremely useful to share the same
language across the team* and that this language be part of the process of
planning, designing, building and researching. I'm not saying that it is
impossible to create Tor Browser without a shared language, nor that there
isn't one already. But I think it could be useful to think how we might
categorize the influence we want to have in users and how they access the
Internet as this is a high-level discussion that is shared throughout the
process by everyone in the team.

*Shared language*

In 2011 *Tromp, Hekkert, and Verbeek* published an article* on "A
Classiﬁcation of Inﬂuence Based on Intended User Experience
<https://direct.mit.edu/desi/article/27/3/3/69045/Design-for-Socially-Responsible-Behavior-A>"
which I think can help us get started and test if this language model
actually works for us. In it, they define two dimensions: force and
salience. *Force* being how strong is the influence of the design to the
user's behaviour. *Salience* being how explicit this is for them.

*Testing the model.* I went back in time to see how we might have used this
shared language in features that were shipped.

For example, before TB 9.0 if the user resized their browser, there was a
small warning advising users how that might expose them, as they could
reveal their fingerprint. So, we are faced with this problem: people are
being tracked because they resize their browser. And they are either not
aware of it or just dismiss it as something absurd to worry about.

How do we assess this feature (the warning)? Using this shared language we
could discuss what type of category it belongs to. I could argue that is a
persuasive influence as people can be aware of it and it's influence is
weak. But then, if users are dismissing such an important warning that
might expose them to trackers, what should we do? Should we have a stronger
influence?

This shared language could be a valuable tool for designers and engineers
to think how might we encourage/discourage certain behaviours and to
imagine beforehand how to assess the impact it would have on the user's
influence. It's clear, looking back, how Letterboxing
<https://blog.torproject.org/new-release-tor-browser-90> is a great
solution for this problem (making the decision for them) or why it was a
good idea to get rid of the Onion Button
<https://securityinabox.org/media/torbrowser-osx-en-v01-323-network-settings-via-onion.png>
(separating it's functionality: making visible the circuit and creating a
New Identity button), but when we are faced with the problem it could be
very messy.

And finally is also useful for research purposes as we can try to
categorize our findings on each feature and build a shared understanding of
what is the actual impact that has in the user's behaviour. This would
inform what's important to work on, it provides context to the team as to
what to do about it and it helps us navigate clearly the decision-making
process.

*Back to the present moment*

Last week I ran a small exploratory study (here's the report
<https://gitlab.torproject.org/tpo/ux/research/-/blob/master/reports/2021/UR-Tor-CostaRica.md>).
In it, two patterns clearly emerged: (i) most participants didn't
understand the main differences of TB with other browsers, (ii) most
participants are not heavy technical users. And so, to try to use the
framework of this language with the findings of the study, I will explain
two specific issues identified.

*First*, it's clear (in the study) that users were not completely aware of
how TB works and thus they didn't understand the increased loading time, or
why they couldn't access certain websites, and when they were asked how
different it's from other browsers they mainly focused on the appearance.

*This is a huge problem*, as our underlying design is decisive: as in,
users must connect to at least 3 relays to work. But is somewhat "hidden"
to users and so they don't understand the implications that this design
choice has in their experience. Thus our challenge becomes much clearer
given these facts: how do we make the design much more apparent so that
users understand the inner workings of Tor Browser? We could think of
different solutions, but just on top of my head:

   -

   show the Tor Circuit in the about page and not show it (only) inside the
   TLS/Certification information.
   -

   show a diagram while they are configuring the browser, or in the actual
   connection screen.

*Last*, another issue that I stumbled upon during the session, is that a
user wanted to change the language of the Browser. And when you try to do
that, it shows this warning
<https://archive.org/download/warning-language-privacy-tor/warning-language-privacy-tor.png>,
which confused the user. Using the model described above, we could
categorize this warning as a coercive influence. As it needs the user's
input to get out of that screen.

Context: This screen has two goals: (i) to know if the user wants to
solicit the English version of websites, (ii) and to warn them that they
could expose their privacy by not doing so.

Without the influence categories we could discuss how we might improve the
writing so that users are not confused by it (and keep it as a coercive
influence). But we could ask ourselves if we should even allow the user to
make this choice and make the decision for them.

   -

   If we decide for them, and make the default option to continue to ask
   for the websites in English (decisive influence), then we would prevent
   users from making a possible mistake and provide them with our best efforts
   to keep their privacy.

We could still make our best arguments regarding if we should make the
decision for them and not even present the choice. But these are
discussions we could all be a part of and the language provides a great
deal of shared understanding within the team and the choice we end up
shipping to users.

--

Sorry about the long email. I'm more than happy to know your opinion
regarding this, either by email or at the next IRC team meeting on
06-04-2021.

Thanks for reading :)

**Tromp, N., P. Hekkert, and P.-P. Verbeek. 2011. “Design for Socially
Responsible Behavior: A Classiﬁcation of Inﬂuence Based on Intended User
Experience.” Design Issues 27 (3): 3–19. doi:10.1162/DESI_a_00087
<https://direct.mit.edu/desi/article/27/3/3/69045/Design-for-Socially-Responsible-Behavior-A>.*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.torproject.org/pipermail/ux/attachments/20210327/5e63c1f6/attachment.htm>