Jakob Nielsen's Alertbox, August 18, 2003:

Mobile Devices: One Generation From Useful

Summary:
New mobile devices show a huge improvement over previous generations, but they're still not good enough to score a real win. To get there, we need both PC-integrated applications and specialized mobile services rather than repurposed website content.

The latest mobile devices are agonizingly close to being practical, but still lack key usability features required for mainstream use.

I've been using a T-Mobile "Sidekick" as my combined PDA and cell phone for the past half year. The Sidekick is also known as the "Danger Device" or the "Hiptop." When I saw an early Danger demo two years ago, I was excited about its potential. Now, after actually using it, I've concluded that one or two more generations of device designs are needed to achieve true usability.

Design Wins

Integrating email, IM, online information, and voice telephony in a single device is a huge win. As I discuss below, the Sidekick is insufficiently integrated, but it's clearly beneficial to have a single PDA with all the main features you need when you're away from the office. Finding people on your contact list is easy, and to call them you simply click a button.

For most mobile applications, deck-of-cards is an appropriate screen size, whether for manipulating an address book or sending a short email. Early dataphone screens were ridiculously small, which made it hard to browse address book entries and read email. Sure, bigger screens would provide even better user interfaces, but the deck-of-cards form factor is big enough to offer good visibility of options and information. It's a more than acceptable compromise, given that we need to carry mobile devices around all day.

For any task involving information, QWERTY keyboards hugely increase usability relative to numeric keypads. Writing email? Of course. But also browsing online information services, or looking up address book entries to place a call.

Design Flaws

The biggest problem with the Sidekick is that it relies on T-Mobile's cellular coverage. There is no polite word in the dictionary to describe the quality of service that T-Mobile coverage offers. Readers from Europe, Asia, and Australia have no idea about such problems, because their cell phone services actually work. Americans know what I am talking about, because most other U.S. service providers are about as bad as T-Mobile. It's common, for example, to be out of coverage inside major airports, where you really need email because you've either been on a plane or are boarding one and will be out of touch for several hours. But airports are hardly the only problem; I've even suffered dropped signal in Silicon Valley.

A low-quality service provider is obviously not a fundamental problem for mobile computing, since better networks can be built. But signal strength is certainly a practical problem, and one that is likely to cause the U.S. to lag behind in the evolution of mobile services.

That said, navigating on the newer devices is much more pleasant than on a WAP phone: you can see more options on each screen and get the next page relatively quickly. Still, page downloads are too slow for extensive interaction. Reading one or two articles is okay, but you wouldn't want to explore a website on a mobile device.

Also, a scroll wheel is an uncomfortable input device for dealing with full-featured user interfaces for two reasons:

A real screen brings a real graphical user interface with a two-dimensional layout of buttons and controls. A one-dimensional controller is awkward for operating GUI widgets.
Online services are incapable of limiting themselves to short pages. Most pages that users download from the Net are long, and scroll forever on a small screen.

Rather than manually scrolling a wheel through every line, we need a 2D input device that moves more easily through large amounts of data. I'm partial to tilt scrolling, where the device senses when you flick it. Bigger physical movements generate faster scrolling, and a double-flick takes you to the top or bottom of the page.

Tighter Integration Needed

Despite its name, the Sidekick doesn't play well with others. In particular, it doesn't integrate with your main computing platform. Yes, you can transfer a copy of your contact database from Outlook to Sidekick, but you can't synchronize the two copies. Nor will updating an appointment on one machine update it on the other. Result: either you have double the work or (more realistically) your mobile device will quickly stop reflecting your actual office situation.

The Sidekick has an always-on Internet connection, and most office PCs do as well. There is no reason why the two couldn't automatically synch in the background without additional work on the user's part.

This lack of synchronization is incomprehensible, especially since easy synchronization was a big design win for the pre-online generation of PDAs: the Palm Pilot synched, the Newton didn't, and almost everybody bought Palm.

Email could also be more tightly integrated with the telephony features. When you receive a message containing a phone number, the device ought to auto-recognize it and make it into a button so you can place the call.

Online Services Must Specialize for Mobile

Email must be reconceptualized for mobile devices. The old model of "anything sent to this address goes into this mailbox" doesn't work for mobile. Either you get too little mail (because you use a special address that nobody knows) or you get too much (because you forward your main address to the device). We need both better filtering and a way to summarize mail that arrives at the office so you can get what you need on the road without being bogged down by a flood of non-urgent messages.

Information browsing also needs to change. Currently, the best we can hope for are websites that are basically scaled-down and redesigned to eliminate graphics and multi-column layouts. At worst, websites offer no mobile version, so you get crunched images and skinny columns that are almost impossible to read.

Clearly, traditional websites are intended for a big-screen user experience. Putting them on a small screen is like the dog that sings: the miracle is that it does so at all. While a technical feat, usability is never going to be good.

To cater to mobile devices, websites and services should offer

much shorter articles,
dramatically simplified navigation, and
highly selective features, retaining only what's needed in a mobile setting.

Right now, there aren't enough mobile devices in use to warrant substantial investments in designing and implementing specialized services, so mobile users have to suffer through repurposed content. Unfortunately, this is a chicken-and-egg problem, because the use of mobile devices will remain limited until more specialized services are available.

To succeed, mobile devices must feel like an extension of your main machine: they must provide what's required, but no more, and add the ability to reach home and grab anything you need but didn't bring. Developers need to rethink applications and base them on a new task analysis that's strongly situated in the mobile context and the moment of use.

Overall, my outlook is positive. The experience of using the Sidekick has proven the potential of mobile data services and integrated PDAs. One or two more generations, and we'll have something good.

Learn More

In-depth course on mobile usability at the Usability Week 2009 conference in Washington DC, San Francisco, London, and Sydney.

> Other Alertbox columns (complete list)
> Sign up for newsletter that will notify you of new Alertboxes