Archive for November, 2009

Redesigning Firefox’s Addons Manager

When I ask Firefox users why they love Firefox, the answer often isn’t because of Firefox. Rather, it’s a particular Firefox add-on that provides functionality that has become invaluable to users. From developer add-ons like Firebug to social add-ons like StumbleUpon, a single add-on can fundamentally change how users interact with the web.

Firefox users get starry-eyed when describing their favorite add-ons

The reason add-ons are so important is because they are a fundamental way that users take control of their online life. Firefox touts its customizability as one of its main selling points, but users’ ability to customize their browsing experience is dependent on the talent and creativity of the add-on developer community. I’ve written in the past about the importance of Firefox providing a user experience ideal for as many people as possible right out of the box, without add-ons installed. But each user is truly unique and uses the web in increasingly different ways. That’s why it’s so important that add-ons be trivial to find, install, and begin using.

The Current Add-ons Manager

Add-ons are currently installed, maintained, and configured via the add-ons manager in Firefox. This window, found under the Tools menu, provides an inventory of installed add-ons and allows users to update, install, remove, enable, and disable them.

The current add-ons manager works, but could use some love

The add-ons manager has been largely unchanged since 2007, and it badly needs a redesign. One reason is that it has several usability problems that would provide significant benefit to users if fixed. For instance, the process of updating add-ons is currently characterized by interrupting important tasks such as startup. Locating a particular installed add-on currently involves navigating through categories such as extensions and plugins. Even experienced users I talked to find the distinction between these categories hazy. A redesign to address current issues should insure that installing and updating add-ons is trivial, notifications are non-disruptive, and the interface provides clear information about the state of a user’s add-ons.

However, a successful redesign of the add-ons manager must not only fix problems, but add functionality. This is because the scope and functionality of add-ons has increased dramatically and will continue to expand in future versions of Firefox. The current add-ons manager gives only the name of an add-on, an icon, a version number, and a one-sentence description. Users could benefit from more information, such as a description or a screenshot showing what UI the add-on will change. Added functionality is also needed because of new ways to modify Firefox, such as Jetpacks and Personas. These are similar to current add-ons in that users can choose items to download for added functionality, but they are installed, managed, and used differently. A redesigned add-ons manager must be able to incorporate emerging projects like these.

Redesign Priorities

From fixing current usability problems to adding needed functionality, there’s a lot that needs to be tackled in the add-ons manager redesign. To help focus the effort, the main areas to address can be described as five priorities:

1. Maintaining and Configuring

  • Allowing users to quickly locate a particular add-on by name or by type
  • Providing simple, usable controls for basic add-on operations
  • Allowing new forms of add-ons, such as Jetpacks and Personas, to be maintained and configured easily alongside traditional add-ons

2. Updating

  • Updating add-ons automatically by default. I’m increasingly convinced that most users, once they’ve decided an add-on is trusted, do not want to manually update it. They especially do not want to be reminded to update when they are starting the browser
  • Allowing users the option to update add-ons manually, update only a particular add-on manually, and possibly to undo an update

3. Installing

  • Streamlining the install process to as few steps as possible
  • Providing the user with clear indications of what action is needed, especially when some add-ons require a restart and some do not

4. Discovering

  • Providing a compelling first-run experience to new add-ons users, including clearly showing what functionality add-ons provide and what they will change
  • Providing a usable, findable way on the add-ons manager to search all existing add-ons, only requiring a visit to AMO when greater community involvement or information is sought

5. Troubleshooting

  • Providing ways to determine if a particular add-on may be causing performance problems, such as ranking by size, CPU, etc
  • Giving clear communication and instructions if there is a security problem with an add-on

This is still very much a working list open to feedback and changes (especially via comments here or on the wiki).  Basically, what I’d like to focus on is making add-on usage less disruptive and more accessible. Experienced and especially technical users tend to be very aware of add-ons and the functionality they provide. These users may see add-ons mentioned in the tech press, may talk to their friends about their favorite add-ons, and might even get involved in the add-on developer community. These are the users who say “I can’t imagine a world without add-on X.” But the benefit of add-ons is felt almost soley by this group. There are thousands of add-ons available that can improve the online experience of just about any user.  Both users and developers deserve an add-ons manager that helps make customizing the browsing experience simple.

How could Microsoft’s Proposed Browser Ballot be More Effective?

(Note: This is my personal opinion and doesn’t reflect Mozilla’s official position nor any formal statement from Mozilla)

In my post on October 15, I wrote about the European Commission (EC)’s investigation of Microsoft due to its bundling of Internet Explorer (IE) with Windows, which the EC viewed as potentially harming consumer choice and innovation on the web. Microsoft, to appease the EC, proposed that Windows users be presented with a ballot in which they could choose which browser to install. I said at the time, and in a subsequent post, that creating a ballot would not successfully address the EC’s concerns nor provide a good experience to users. However, since the EC seems to be giving Microsoft the go-ahead to design a ballot, it seems that the best we can do is consider how to design a ballot that causes the least amount of harm to users.

Here are two broad principles that are important in ballot design.

Principle 1: A ballot should be clear and simple

A ballot should present the voter with the information they need to make an informed choice, but no more. The verbal and graphic language on the ballot should be organized so that readers follow a consistent path through the ballot’s information.  Many viewing the browser ballot will not be familiar with the browser as a separable element from the operating system, so clear and precise language is vital.

In interaction design, the complexity of a new task can be lessened by leveraging against what the user already knows and expects. For instance, it’s usually best to display the instructions in the top left for western readers, and before the the possible ballot choices (Kimball and Kropf 2002). It is also recommended that the instructions be as close to the first task as possible, because it provides the highest chance that voters will read them (Dillman and Christian 2002 (pdf)).

Other recommendations to make a ballot simple and clear include:

  • Instructions written short and simply, and in an active, affirmative style (Sanders and McCormick 1993)
  • No unnecessary information or clutter around the choices (Niemi and Herrnson 2003)
  • No text smaller than 12pt (Roth 1994), left justification preferred (Dillman 1978), shading and highlighting used to direct the voter’s focus (Kimball and Kropf 2002)
  • Ballot items listed in a single row or column (Darcy and Schneider, 1989). Failure to do this is part of what caused the ballot problems in Florida during the 2004 US Presidential election
  • Enough spacing between lines to highlight groupings of visual elements (Dillman 2000)
  • No ambiguity over which button corresponds to which choice (Kimball and Kropf 2002)

Principle 2: A ballot should be impartial

I wrote in my previous post about how ballot order can influence voters. Another important factor is how much physical space on the ballot each item is designated.

The current design runs into some problems with designated space per item, swayed in favor of IE. For instance, in the current design:

  • The user must double-click on an Internet Explorer icon, labeled “Internet Explorer”, to launch the ballot
  • The ballot appears within Internet Explorer browser chrome
  • Internet Explorer is mentioned repeatedly within the ballot, Bing is shown as the default search engine, and the IE logo appears as a favicon multiple times

The current space allocation for IE is roughly 3.35 times as much as the other browsers.


Current ballot: 3.35 times more space given to IE than the other browsers


Allocation of visual space for each ballot choice

This space that IE occupies in the top left of the ballot is particularly important because it’s where users begin an eye scan of a new page.  As an anonymous source pointed out to me, Microsoft mentions this in their layout guidelines:

“All things being equal, users first look in the upper left corner of a window, scan across the page, and end their scan in the lower right corner.  They tend to ignore the lower left corner.

But in interactive UI, not all things are equal so different UI elements receive different levels of attention. Users tend to look at interactive controls—especially controls in the upper left and center of the window—and prominent text first”

Improving the ballot based on these principles

In summary, the ballot would be improved by being simpler, clearer, and be presented with equal weight to each browser.

Here’s a version similar to the current proposal but with these principles in mind: