The first thing that comes to mind when I think about my home audio and video equipment is the ever-present blinking “12:00″ on our VCRs (now DVD players I suppose).

I was very much surprised to find a terrific example of usability in a niche product for those of us with too few component video and/or digital optical inputs: the Audio Authority 1154A. It’s the perfect example of something I wouldn’t have asked for but is exactly what I needed; goal-directed design at its best!

I’ve got too many devices and not enough inputs. Here I am updating the system software on the Playstation3 (right-most device on the top shelf).

Many “serious” A/V switchers can be remotely controlled as home theater enthusiasts typically sport universal remotes as part of their gadget repertoire. The remote control commands for all devices are added to a single device that rules them all. If your target market has a remote, makes sense to give them what they want right? An A/V switcher with a remote?

Not quite.

My goal when sitting down isn’t to mess around with my gadgets, it’s to have fun playing Rock Band! Enter the 1154A, the perfect goal-driven device:

No remote and only one button. It automatically routes the signal based on which device is powered up (my PS3 in this case). And as turning on a device is a prerequisite for using it, this switcher is completely invisible.

Multiple devices turned on? No problem, it gives priority to the higher numbered device (connections in the back are numbered) - simple as that.

Really phenomenal design here and not at all something I would expect from the home theater enthusiast industry which is typically more about the size and complexity of your universal remote than how easy it is to use. Best $250 I’ve spent in recent memory even though I sometimes forget I own it :)

UPDATE: Hopped in the shower after posting this when it hit me - look at where the 1154A is on my setup.  Top shelf, facing forward,

as though I need line-of-sight to use an infrared remote with it

This 1154A could be behind the receiver, out of sight, out of mind.  Instead, I’ve positioned it as though I’m using a remote with it, even though I’m not.  Old habits…

While anxiously forwarding iPhone photos of my Expert drumming skills on the new Rock Band Boston DLC, I noticed something:

What surprised me was the inclusion of the MIME type adjacent the attachment name and size. Actually a few things were surprising…

Let’s just take on this entire part of the interface:

• The entire attachment doesn’t need to be a link, as the link should target the canonical form of the content - the attachment name itself. Who is going to click on the size of the attachment to look at it? Increasing the size of the click target isn’t an issue here - attachment names are of generally sufficient size that this isn’t an issue.
• The MIME type is obviously excessive.
• I’ll go one step further and say that the attachment size isn’t important either. If the user won’t modify their behaviour based on the size of the attachment, why bother displaying it?

If Google’s objective is to translate the meaning of the .JPG extension, why not really commit and just call it an image? And rather than report the size, why not abstract that away as well?

Now we can concentrate on the actual attachments instead of the metadata.

As web applications become more interactive, there has been a tendency to turn them into their desktop brethren.  As a simple example, the close button as it appears in Microsoft Outlook Web Access 2007.

This is akin to using the MDI (Multiple Document Interface) which Microsoft promoted and disowned years ago.  This was the idea that a window could contain multiple windows - a desktop within a desktop; a usability nightmare.

Moving on, the most notable point of comparison is the dialog.  Yahoo’s YUI even makes creating them easy to do - draggable and everything.

The most notable offender in my day-to-day experience is the strangely named “Rally Program” by Rally Software Development.  Let’s say you’re working on a User Story and you’d like to change one of the many bits of associated data, for example the state or rank.

In order to change them I have to seek out the Edit function (virtually hidden in a pull down).  Here we have a small section of the Edit dialog:

One is left wondering why the extra step is necessary.  If we imagine dialogs as some suggest (i.e. “separate rooms”) this becomes a clear form of excise.

My father used to work for the PGA Tour and once met with Ron Jaworski.  At the time I knew of him as the backup quarterback for Dan Marino.  I remember him saying “If you can touch it, you can catch it.”  I’m sure his wide receivers heard that a lot :)

Likewise I have a similar saying for information in applications - if you can see it, you can edit it.

Rally Program does feature inline editing in some instances, not nearly enough however.  Fortunately Rally Software has a very friendly and responsive product team so I have no doubt things will improve in the future.

From our VP of Marketing:

What’s most interesting to me is the focus not on WebEx as it relates to our internal usage, but how our customers relate to us through WebEx.

Have you considered how your customers’ users’ customers (i.e. Cooper’s “served personas”) experience your product and what that might say about your customers? Recall how the above email ends:

” …is much more aligned with the usability and simplicity that customers expect from [us].”

This speaks to design on Norman’s “reflective” level as discussed in his book Emotional Design, paraphrased below from Reimann (a contributor to About Face) on UXmatters:

“reflective—The least immediate level of processing, which involves conscious consideration and reflection on past experiences. [snip]. The most interesting aspect of reflective processing as it relates to design is that, through reflection, we are able to integrate our experiences with designed artifacts into our broader life experiences and, over time, associate meaning and value with the artifacts themselves.”

Essentially, what does this say about me and/or my company?

Creative Labs learned this lesson the hard way. Apple threw concern for cost and features to the wind and took over the MP3 market that Creative basically created with their Nomad line of players.

Are your customers in danger of leaving for this reason?

How convinced are you?

My first experience with Hellgate: London was an error dialog.

“Could not find language.dat”

A quick visit over to the Flagship Studios’ forums (HGL developers) only to find that it’s my vault.

The relevant section:

“…is the result of you people installing through [the wrong file]…you must run the setup.exe on the DVD”

setup.exe? Can you find it? Quick! You hurriedly unpack your shiny new game and click on the first thing you see that looks reasonable.

The “HGL_x86 Windows Installer Package” looks good to me - I have Hellgate, I’m running on an x86 CPU and I want to install it. The tooltip confirms my line of thinking. The install proceeds without warning, only the game won’t start. D’oh!

There is a simple solution here - place the installer packages in a subdirectory. If the contents of the DVD looked like this:

What do you think the chances were of me making this mistake?  With only one real choice, little to none :)

When deciding whether or not to use a link or a button, I have one simple rule - links are for going, buttons are for doing.  Take this example from the Microsoft Windows Vista photo printing wizard.

While I may have some improvements in mind for the layout, the usage of standard (or now standard) components is spot-on.  Buttons are used to alter the flow (continue or exit) whereas links are used for navigation.

The links bring up additional configuration dialogs that alter the print operation but keep you in the flow.  While this has the potential for distracting you from the flow with multiple nested dialogs the likelihood of usage detracts from that concern.

Concern should be expressed at the over-usage of links in desktop applications but like any new component, designers will get a better feel for them over time.

I have mixed feelings on the usability of eBay in general. They’ve optimized the selling flow and rightly so as they make all of their money on listing and final value fees. Twice this week it occurred to me that I should be listing additional items in my early spring cleaning. Selling is so terrifically easy that between meetings I was able to list both items in about 3-5 minutes.

In the past I’ve pointed out the over-abundance of controls in the mail interface. Today I want to talk a bit about the modification of navigation elements in that same interface.

As a seller, I often perform two operations:

1. Check the total selling price of all of my items (the single green number on top of my “Selling” page).
2. Respond to messages (i.e. questions from potential buyers, payment-related issues after items have closed).

Both of which are easily accessible from the left-aligned navigational structure in the “My eBay” section of the website.

As you can see I’ve sold 9 items with 1 pending and while authoring this post, another message has just arrived. Let’s respond to that message.

Note that my left navigation is gone even though the breadcrumb indicates that I’m still in “My eBay”. Normally I’d just hit the Back button. Taken out of context, that would be my analysis as well. Typically, however, I respond to messages which means several presses of the Back button to get back to My eBay. The problem with this approach is that I frequently “overshoot” my target: the “Sold” items page.

How about using the “My eBay” quick links in the top right?

Nope, can’t get back to sold items from there. Note my confusion, as indicated by the question mark. Thus begins the multi-step process of figuring out which of these links will take me closest to the Sold page (all of them, it turns out will put it one click away).

This “reeks” of separate teams designing the interface, or at least of the messaging interface being designed in seclusion without a holistic view of the selling and settlement processes.

A dialog from the Cisco VPN installer reminded me of an episode of Seinfeld.

Seinfeld Season 9 - “The Maid”

JERRY: What’s around you?

KRAMER: I’m lookin’ at Ray’s Pizza. You know where that is?

JERRY: Is it Famous Ray’s?

KRAMER: No. It’s Original Ray’s.

JERRY: Famous Original Ray’s?

KRAMER: It’s just Original, Jerry!

JERRY: Well, what street are you on?

KRAMER: Hey, I’m on first and first. How can the same street intersect with itself? I must be at the nexus of the universe!

Cisco’s VPN software is/was notorious for its stability issues on the Mac. Of course skepticism followed me to my Vista laptop. Every so often it would forget all of my connection settings forcing me to re-enter several bits of easily-forgettable information: IP addresses, obscure passwords and VPN configuration settings.

Let’s get the latest version.

Start the installer:

This is an upgrade from a previous version yet the installer is asking me to uninstall… itself? I must be at the nexus of the universe!

Consider the following:

1. Cisco was able to detect this condition.
2. Cisco wrote the installer for this application.
3. Cisco wrote the installers for all previous versions of this application.

Perhaps Cisco, with its intimate knowledge of its own software, might be able to handle all of this for me? Isn’t this what computers are great at? Automation?

“Aww c’mon Rob!” you’re thinking.  “Just uninstall and reinstall yourself.  It can’t be that hard!”  Here’s how that process went:

1. Uninstalled original client which crashed the machine.
2. Uninstalled (again) ’successfully’.  Reboot.
3. Installed new client, detected remnants of previous install (even though I uninstalled it), rolled back install progress, started some sort of Cisco uninstall.  Reboot.
4. Installed new client (again).  Reboot.
5. Installed new client (again for the third time) and was successful.

Let’s learn from Cisco and Yahoo!’s less-than-stellar examples of error reporting - if we can detect it, we can probably do something about it.

McAfee’s update notification configuration dialog has several options for us to choose from. Unfortunately they all suffer from the same fundamental problem:

Each of them requires user intervention.

The less aware users are of our applications, the better. This is difficult for many from both sides of the aisle to internalize (business, engineering) and I believe it is primarily a psychological concern - many people lack the courage of their convictions that their applications should be instilled with.

A security update is available? Download and install it! Big download? Chunk it and do it in the background. Requires a reboot? That’s fine, the machine will be rebooted eventually. In the rare case where it’s a code red, a prompt is acceptable - those situations are few and far between though.

The normal mode of operation for our applications should be not be to “stop the proceedings with idiocy” (to quote Alan Cooper) but to proceed as one would expect.

Borrowing from Fowler’s “Refactoring,” I’m using the phrase “bad usability smell” as something that tips us off; that things might not be quite right.

In order to return my notebook warranty, HP provided a link to an online FedEx label printer.

Notice the two bits of instructional text: “(Please select one or more labels)” and “Click the Continue button for dropoff/pickup options.” Why is this a bad smell? Such a simple interface should not require instructions. It doesn’t use any complex components. Checkboxes and buttons have been in the mainsteam for almost 20 years. We have:

• (2) Buttons
• (1) Checkbox
• (2) Links - We can ignore links since navigability is implied, they should be ignored by users as well.

What’s wrong?

1. The “Print Selected Labels” button is out of sequence. If the first step is to “select one or more labels” it should be after the checkbox(es).
2. The enabled/disabled state of “Print Selected Labels” is all but impossible to discern.  By disabling it, they’re giving you a hint that a requirement isn’t met, but by such obfuscation there’s no way you’ll be able to tell when it’s disabled.

   

3. The “Continue” button should be a link. It’s acting just like a link, moving you on to the next step - a navigation function.
4. The “C” section heading is in the same font size as the table headings.  If the function is different (section label v. column header) it should look different.
5. Arial and Times fonts are mixed - why?

Instructional text usually tells us that the text itself should be incorporated into the interface.  Quick mockup:

There’s much room for improvement and I’m late for work.  Hopefully someone else picks it up and runs with it :)