“My Life” by Milosh, turn it on, and turn it up.

When was the last time you saw a floppy disc?

Unless you’ve recently accomplished some spring cleaning in your office, my guess is that it has been too long to remember the last time you saw a floppy disc, and arguable that it has been over a decade since you have used one.

Then why is it still common for designers to use an icon of a floppy disk to indicate the function of ‘Save’ even though floppy disks have not been used since Y2K?

‘Save’ is Dead

It has been discussed amongst my fellow Mad*Power’s that “saving” is no longer an isolated action, and that the lines between saving and sharing have been increasingly blurred. Rather than storing information into the depths of your hard drive, behaviors and technology have evolved to the point where we no longer want to strictly ‘save’ a file; we would rather post it, tweet it, upload it, text it, or email it. With the addition of Apple’s “Time Machine”, which saves everything automatically, it is implied that the need for a user to proactively protect their files will soon be obsolete.

The question remains, is the floppy disc still a relevant icon? When Mad*Power’s were challenged to find a new icon that represented ‘save’, the following icons below were chosen, and illustrated how a single word can have a variety of interpretations:



Iconography what?

The word iconography means “image writing.” In other words, to be able to suggest to the viewer, what something means without having to write it. A great icon will suggest to the user the primary purpose of that function without having to read accompanying text. The very best icon will accomplish the above and be instantly and universally recognizable.

Why then, when given the task of selecting a “save” icon, did Mad Power’s select so many different icons?
It’s important to remember that words can have many meanings. Save, for example, can also mean secure, savior, savings, conserve, storage, rescue, etc. Each person, therefore, selected the icon that best fit “their” interpretation of the word “save.”

Outdated? Who Cares!

With so many icons to choose from, how is it that everyone still manages to navigate our complex iconographic world?

The answer is universal association.

Does it matter if the floppy disk icon is outdated and untrendy? My answer is no. People have a strong understanding and universal association of what is implied by that disk. This is where the rule of grandfathered icons comes in.
Some icons aren’t accurate or current in their depiction, but still have a strong understanding and universal association for a particular function or meaning. For example, what about these “grandfathered” icons?

•    Suburban house  » Home
•    Envelope  » Email
•    Eraser/Pencil  » Edit
•    Magnifying glass  » Zoom
•    Light bulb  » Idea
•    Trash can  » Delete
•    Paper clip  » Attach
•    Heart  » Bookmark (Favorite)
•    Head silhouette  » Profile
•    Gear cog  » Settings
•    Zipper  » Compression

My ultimate opinion is, use whatever (floppy disk or not) resonates best with the broadest set of users, knowing that ultimately, if the user has to think about an icon to 'get it', the metaphor is too complex or irrelevant.

Recently, Skype redesigned it's IM client for Mac. Like many users, I was excited for the expected enhancements that usually come with new application releases.

And, like many Mac users, I was disappointed with what was actually delivered. (apoplectic might more appropriately describe my Mac brethren).



Blindly fanning the flames, instead of fixing the perceived issues, Skype decided to hold a contest to let someone else fix them.

For the last couple of weeks now, Skype-hate has run rampant all over the web. The user comments at the end of this blog post make for a nice overview.

Amidst the uproar, I wanted to chime in and point out what a great teaching tool this mess can be for user experience designers everywhere.

Lesson #1: Talk to your users (yes, I realize it's a frighteningly easy one).

I have no proof that Skype didn't talk to its users, but judging from the feedback, I'm guessing they didn't. Asking their users would have revealed some fairly simple needs:

• Keep my IM application small
• Allow me to configure what I'm looking at
• Don't change a ton, I like it the way it is

ISO-9241 states that usability is “the effectiveness, efficiency, and satisfaction with which specified users achieve specified goals in particular environments.” Some of the articles that I’ve come across take an overly simplistic view of usability research and could guide the uninitiated into treacherous waters. These state each of the three measures of usability (effectiveness, efficiency and satisfaction) can be directly measured to achieve quantitative results in any type of usability study. To measure effectiveness, just keep track if the users pass or fail the tasks. For efficiency, time how long it takes users to complete the tasks. Finally, for satisfaction, simply ask the users how satisfied they were on a 1-5 scale.

Folks, this is bad advice for a number of reasons. First, due to budget constraints and diminished returns, a moderated usability study typically has 8-15 participants. This is nowhere close to enough subjects to get statistically significant data. Sure, we may get some skewed data that will give us useful information, but 8-15 people are certainly not enough to perform true statistical analysis. If a researcher tells you that s/he interviewed 15 people and can therefore predict what the population at large will do, find yourself another researcher. The rule of thumb is 30 participants for the possibility of statistical significance. It all depends on the variability of the data. If the data is all over the place, you’ll need more than 30 participants.

The second reason this is bad advice is that the data collection for these types of data points can be quite subjective. Consider pass/fail for a moment. What exactly constitutes a pass? Is it when the user thinks s/he is done or when the researcher notices that the user clicked the link that was predetermined to be a pass? Anyone who has run a number of usability studies has seen users successfully complete the task, but indicate that they have not done so.

Now consider the efficiency metric: how long it takes a user to complete a task. If you are doing a think-aloud protocol, where the user is asked to talk through the task, just throw away the efficiency metric. Since the moderator will be conversing with the user, it will add time to the task and make the metric useless. Any time someone is in the room with the user, the efficiency metric most likely won’t work. Even if you don’t specifically ask the users to think-aloud, they may ask the moderator questions or otherwise delay finishing the task.

Finally, let’s consider satisfaction. This one is nebulous at best. How exactly do you ask a participant if the task was satisfying? If you ask, “how satisfying was that task?” after each task, the participant will get habituated to the question and give inaccurate responses. However, there are a few standard questionnaires that can be used to gather this data: System Usability Scale (SUS), Software Usability Measurement Inventory (SUMI) and the Wong-Baker FACES Pain Rating Scale. But, this brings us back to the issue of statistical significance. If you use one of these scales in a moderated usability study, the findings will be directional, not statistical.

Don’t get me wrong, there is a time and place for asking these types of questions: in an asynchronous remote usability study with many participants. In an asynchronous study, researchers use services like Loop11 or UserZoom, which facilitate a high number of participants through a self-guided usability study. The tasks and questions appear at the top of the page, with the content just below. This format alleviates the problems that I have mentioned. A large number of users can participate in the study, allowing at least the possibility of attaining statistical significance in the data. For the effectiveness measure, users typically press a button at the top to indicate that they have completed the task or give up. Since we know what links the user has clicked, we can determine if s/he correctly completed the task.

For efficiency, time on task is an accurate measure in an unmoderated study. The system will keep track of the time it takes the user to complete the task and the user will not be distracted by a moderator in the room. This isn’t to say that the user won’t get otherwise distracted, but those outliers should be fairly obvious. As for satisfaction, the high number of participants should allow statistical significance in the data.

The moral of the story is: use the right tool for the right job. If you are running a moderated usability study, you aren’t going to get statistically significant data. As such, you must ask questions that rely on qualitative responses so that trends can be found among the participants. Yes, you can ask quantitative questions, but these results will be directional at best. If you have a client who insists on statistical significance, or just wants a larger pool of participants, consider performing an unmoderated asynchronous usability study. Don’t fall into the trap of calling data a statistic, when it really isn’t!