Jakob Nielsen's Alertbox, September 15, 2003:

Time to Make Tech Work

Summary:
The IT industry is maturing. Hopefully, this maturity will result in a slower introduction of new features, which in turn will let companies focus their attention and resources on making existing technology work better for users.

Information technology is maturing -- a fact that is causing much whining in Silicon Valley. As profit margins narrow in both hardware and software, companies are more likely to commoditize software development and move it to cheaper locations, causing some concern due to the poor state of usability practices in the main offshore countries.

Such decisions might improve an IT company's bottom line, but it's unlikely to improve the unrest among IT customers. The CIOs of the world are getting tired of paying for enterprise software that is so complicated to implement and sustain that the total cost of ownership usually far exceeds potential savings. And the CEOs of the world are tired of funding incessant new technology purchases rather than having IT departments focus on improving employee productivity and making existing features work.

Making IT Work

Earlier this year, Harvard Business Review famously pronounced that "IT Doesn't Matter" to a company's competitiveness. It's certainly true that in the past, computer technology investments were negatively correlated with profitability, and that we had reduced productivity growth for many years in the United States and other advanced countries. The main cause in both cases was an overly rapid introduction of new technology that was unsuitable for humans.

Recently, investments in new technology have slowed and productivity has surged because companies are focused on making existing technology work rather than chasing the latest fads.

My short wish list for making technology work:

• No bugs. Malfunctions cause users to question their system understanding and form erroneous mental models.
• Security as default. Users shouldn't have to install security features and mess with signature schemes; this approach is guaranteed to leave most systems unprotected. Also, no email should exist in an unencrypted state unless it's being read or written.
• Integration. Plug-and-play must become a reality, not just for hardware but also for software and services.
• Reliable wireless. (This point mainly applies to the U.S., since most other countries have wireless services that work.) Having a steady signal that's always there is more important than fancy 3G features. We also need integration between wide-area cellular coverage, short-range WiFi, and other services that let users hand off from one service to the next. These hand offs should always be to the cheapest and best connection, and should not require users to log in or establish accounts with each carrier.
• Multi-device user experience. Users have more than one device and move between multiple locations; the network is the user experience. Synchronization should occur automatically through the network and not require any explicit user action.

Each of these criteria will require an army of programmers to implement. Fine. The result will be far more useful to customers than spending the same resources on a flood of new features that they typically don't need.

Some Innovation Is Needed

Just as Francis Fukuyama was wrong in predicting "the end of history," it would be wrong to predict the end of IT innovation. We still need some new features; hopefully they'll emerge more slowly than in the past, and be more solidly designed and implemented before they're unleashed.

Most importantly, developers must realize that the current personal computer paradigm is coming to an end.

• We need a new type of operating system that helps users manage their time and protects them from information overload.
• File systems and local search must change to accommodate the massive number of information objects that users accumulate during a lifetime of computer use. (The "my documents" folder on my PC, for example, contains 66,794 files in 2,039 subfolders.) We also need to unify two information spaces that are currently separate: documents and email.
• Email must be reconceptualized from its "the mail shall get through" approach to one that focuses on true communication and collaboration (probably offloading most collaboration features to other interfaces).
• We need an Internet control panel to collect notifications and monitor services and events of interest to the user. Current applications like RSS aggregators and eBay auction monitors are early attempts, but we probably need a single integrated control panel.
• Content creation applications must abandon office automation and stop treating printed output like the Holy Grail. Instead, they must become intranet communication managers. (Of course, using the plural term "content creation applications" is a fantasy -- we all know that MS Office is the only one that matters.)

Microsoft's forthcoming attentional user interface is supposed to support many of these changes. Unfortunately, the move from GUI to AUI is not likely to happen anytime soon, particularly since Microsoft is notorious for needing to get to version 3 before a new product works.

Web Browsers: The Road Ahead

Web browsing sorely needs improvement. I've long been predicting that Internet Explorer 8.0 would be the first good Web browser, and it's certainly possible that Microsoft's promised search tool will achieve the long-awaited integration between Web search and client-side navigation. For example, users need the ability to scope a search to consider only "pages I have seen" or "websites I have visited" as well as the ability to have navigation links reveal the relevancy of destination pages for the user's current query. (A recent study at the Palo Alto Research Center found that adding this latter feature almost doubles users' performance: locating products on the Xerox website took 3.5 minutes with plain browsing, 3.0 minutes with regular search, and 1.6 minutes when links were highlighted with search relevance in an experimental "ScentTrails" interface.)

Features like site maps, breadcrumbs, and other structural elements must become browser commands. Extracting information space navigation from the website will liberate users from the whims of Web designers and ensure consistent and standardized navigation features.

The Web's history has shown that people use generic commands like the Back button far more than intermittent features, which, when they are available -- and users can actually find them -- tend to look and work differently on different sites.

Clearly, we need both some new features and a paradigm shift in operating systems. Overall, though, it would be better for users if the IT industry took a collective deep breath, slowed down, and focused on making existing features work.