wiki:doc/TorLauncherUX2016

Version 18 (modified by lnl, 3 years ago) (diff)

--

Overview

Timeline

  • Phase 1: Oct-Dec 2015: User Experience Research
  • Phase 2: Dec 2015-Jan 2016: Iterative Interface Designs
  • Phase 3: Feb - March 2016: User Acceptance Testing

Team

  • Linda Naeun Lee: User Experience and Security Researcher,
  • David Fifield: Security Researcher
  • Nathan Malkin: User Experience and Security Researcher
  • Ganesh Iyer: User Interface Designer

+ Nima, Isabel, and everyone else who has been generously giving us feedback along the way. Thank you!

User Experience Research

Takeaways:

  1. (under construction)
  2. (under construction)
  3. (under construction)

Artifacts:

Qualitative User Experience Research

Most representative behaviors:

  • P3: this participant configured a bridge successfully (obfs3, the default), even though they didn't need to. They didn't know the difference between censored websites and censored internet connections.
  • P11: this participant tries to connect using vanilla tor by clicking connect, which doesn't work against a censor that blocks public Tor relays. Then they try to connect using vanilla tor because they answered that they didn't need or a proxy, which also didn't work. They try connecting with a bridge (obfs3, default) and no proxy, which works.
  • P14: watch this poor user try the same configuration 7 times(?!!) and fail and guess proxy settings. 27 minutes, later, we cut them off. Because they are convinced they need a proxy, they failed to connect because they never bother to pick a different bridge (chose obfs3, required meek or custom).

Linda's favorite 3:

  • P1: this participant makes a correct configuration (obfs3, no proxy), but since bootstrapping took too long, was convinced it was wrong. Then tried "connecting directly" by pressing connect on the first screen, which worked. But the participant was convinced that they made a direct connection.
  • P4: after successful configuration, the participant is convinced that the file explorer on windows IS the Tor browser for a long time, before figuring out what the browser is.
  • P5: watch this participant create an email account on the fly and use the auto responder to use a custom bridge to connect successfully.

Methodology

Experimental Setup

Instrumentation

Results

Interface Designs

Clickable Prototypes

Note: these clickable prototypes are rough, so there may be some missing features (i.e. if you click "help," nothing happens and there is no help window, but that's because we don't plan on changing that, nor did we take the time to re-draft it up). Additionally, there is only one version of each window in the prototype, so the state may seem inconsistent (i.e. in Iteration 4, no matter what bridge you choose, you end up choosing the same one). These were mainly to demonstrate the design changes that we want to make to the interface, so please bear with us! It is quite cool that you can click through the prototypes and explore the paths yourselves.

Tip: Marvel, the site where our clickable prototypes are hosted, will act as a prototype. You can explore the paths manually to iterate through all the windows, but if you want to see all the windows at once and navigate to any window, you can click this icon at the top right corner. It will then display a grid of each window of the prototype. Unfortunately, I can't find a way to display all windows at once bigger than the grid appears. But clicking on a window's thumbnail will navigate you directly to that window, without you having to traverse through the interface path yourself.

No image "marvel-navi-tip.png" attached to doc/TorLauncherUX2016

Changes from Iteration 1 -> Iteration 2

Changes from Iteration 2 -> Iteration 3

Changes from Iteration 3 -> Iteration 4

Changes from Iteration 4 -> Iteration 5

We ran a focus group at Berkeley during one of the BLUES (Berkeley Laboratory for Usable and Experimental Security) Lab meetings. 14 researchers, ranging from a spectrum of computer science to pure user experience researchers were in attendance and provided feedback on the design. The discussion started with an overview of various censorship environments around the world, followed by a discussion of how the 3 simulated censorship environments in the lab were representative of these conditions, a walkthrough of Tor's current (5.0.3) version of the launcher and the ideal path for a user in each of the 3 simulated censorship environments would take, and the results of the qualitative user experience research. Then, we presented our new design (Iteration 4), stated our justifications for changes made from the original launcher to our new design, and concluded with a discussion of the design decisions made and possible improvements.

We believed that sanity-checking our design before large-scale testing would be wise, since it is time-wise and financially costly to run such an experiment. During this focus group, the researchers worked out little kinks in the design, discussed alternatives and made a collective decisions to implement or not implement possible changes, and polished the design to look more aesthetically pleasing. Yay! Here are the changes that will be made from Iteration 4 to Iteration 5:

  1. Auto-proxy detection:
  2. Smarter redirections:
  3. Proxy configuration before Bridge configuration:
  4. Eliminating enumeration:
  5. Visual guidance:
  6. Updated text:
  7. Finalized windows:

(If you want to know more about specific design decision which seemed like good ideas, but we decided against implementing them for various reasons, you can look in the discussion section of this wiki.)

We hope that Iteration 5 will be the last iteration of the user interface. We're ready to get testing on the impact of these changes on real users!

User Acceptance Testing

Coming Feb/March 2016.

Methodology

Experimental Setup

Instrumentation

Results

Discussion

  • Q1
  • Q2
  • Q3

Impact to Tor's Launcher

Attachments (7)

Download all attachments as: .zip