August 24, 1999, Volume 2 - Number 12
With the Web, user demands for interface features shoot to the top of the priority list
The Second Revolution of User Interfaces
Ralph Kimball
Data warehousing is increasingly being delivered through the Web medium. And, conversely, the
Web medium is increasingly inviting the data warehouse to share its treasure trove of high-quality data,
in the same way that other kinds of data are already shared on the Web. As a result, mainline data
warehousing is now inexorably linked with that which stands between the data store and the human source
and destination of the data: the Web interface.
In this article, the 50th in the series I have written for DBMS magazine and, now, Intelligent
Enterprise, I return to my roots as a user interface designer. Twenty-five years ago, when I was
working at Xerox PARC, I had the great fortune to participate in the birth of the modern computer user
interface.
Inspired by the demonstrations of Alan Kay and his Learning Research Group at Xerox, a
whole generation of user interfaces based on bitmapped displays, mice, windows, and icons flowed forth in
a great burst of creativity from PARC in the early 1970s. My job, in the Systems Development Department
of Xerox, was as part of a team that adapted and refined this creative work from PARC and made actual
products. Our baby, the Xerox Star workstation, launched in 1981, was the first commercial product to
employ mice, windows, and icons. Apple’s Lisa and Macintosh, and Microsoft’s Windows all came
later and, like the Star, were based on the demonstrations performed at Xerox PARC.
Of course, the real cognoscenti among you know that even PARC’s ideas had their origins in yet
earlier creative work by Doug Englebart at Stanford Research Institute and Ivan Sutherland at the
University of Utah in the 1960s. But that’s a story that will have to wait until another
time…
The story in 1999 is that we are poised to undergo a second major revolution in user interface design.
This second revolution, propelled by the inexorable forces of the Web, builds on the first revolution,
but defines the user interface in ways not envisioned by the designers of the 1960s and ’70s. The
goal for user interfaces we’re building as we turn the millennium is no longer to make the
computer useful. Our new goal is to make the Web useful.
How the Second Revolution Differs from the First
The first revolution was based on a new medium that allowed a personal relationship with the computer.
This new medium was called, appropriately, the “personal computer,” or “PC.” Although
early PCs had character displays, the revolution really caught fire when the personal computer became
equipped with a bitmap screen, a mouse, windows, and icons. This new interface allowed WYSIWYG
(“what you see is what you get”). It emphasized recognizing rather than remembering, and
pointing rather than typing.
The second revolution is based on an even newer medium; one that allows people access to the services of
the Web. Although early uses of the Internet were based on character interfaces, the revolution really
caught fire with the adoption of a standard format for hypertext links that allowed the development of a
vast interconnection of information consisting of text, graphics, and other media. The impact and the
importance of the Web is difficult to overstate. It is a force like a tidal wave. Within the next few
years, most of the human beings on the planet will probably have some kind of access to the Web. It is
the great homogenizer, the great communicator — and it is also chaos.
The new Web user interface no longer promotes WYSIWYG; it promotes IWIN (“I want it now”). It
is based on gathering information, recognizing the choices that one expects, and getting the results
instantly.
The User Interface Is Now More Urgent
During the first user interface revolution, we all marveled at the PC, but strangely, we have not put
effective pressure on the user interfaces to improve. The computer marketplace still has not provided a
good feedback loop for improving user interface design. User needs and dissatisfaction have not been
communicated directly to the developers responsible for the design. Product evolution has been driven
either by marketing, which wants more features, or by development, which wants more robust
infrastructure. How often have we waited a complete release cycle during which a computer product was
being “rearchitected” without feature additions? Lost in all of this have been the users’
concerns. Advances in usability can come only from a thousand tiny improvements. Until now, when each of
the improvements needed to compete in the vendor’s annual release planning meeting, they often
didn’t have a powerful spokesperson or a critical mass of visible supporters.
Although for the moment the Web is delivered through the first generation user interfaces, make no
mistake about it — this medium is new. The rules are changing, and most of these changes never could
have happened without the constantly pushing force of the Web. The push is real and much more urgent for
the following reasons.
User interface feedback from the Web is personal. Web logs allow us to see gestures of the
individual customer. We usually know who the customer is, what the customer was trying to do on the Web,
and whether that customer was successful. Developers in the first-generation PC world could go their
entire careers without ever analyzing user protocols.
User interface feedback from the Web is immediate. Headquarters can detect individual Web-user
sessions’ success or failure within seconds and analyze the sessions at leisure by the next day. In
the personal computer world, user gripes and usability issues could take a year to accumulate, only to
have to survive the political, release-planning process.
User interface effectiveness on the Web is now tied directly to profit. The user interface is no
longer the product, it is the portal to the product. By the time the customer used a product on a
personal computer, the producer had received payment for it. Now, on the Web, the user interface stands
between the customer and the purchase. Corporate revenue may be tied directly to the effectiveness of the
user interface. We are in the unprecedented situation where the CEO is pounding the table and demanding a
better user interface!
Second-Generation User-Interface Guidelines
Having made the case for better user interfaces, is that all there is? What is different about what the
Web demands from the user interface? A list of user interface criteria has emerged as a kind of standard
checklist for Web user-interface design. (See page 56.)It’s striking that, in light of the fact
these items are so important, none of them found a top position in the first-generation user-interface
guideline lists.
Just how relevant are all these user interface concerns to data warehousing? Client/server architectures
and product development cultures that ignore the Web are dead as doornails. In the next issue, I’ll
take each of the guidelines developed in this article and apply them specifically to the mainline data
warehouse activities of ad hoc querying, high-end reporting, and data mining.
INFO NATION
User Interface Criteria |
|
Near-Instantaneous Performance. Ten seconds is a long time on a Web page. A
page has to deliver the first screenful of its useful content within 10 seconds.
Expected Choices. Every page has the responsibility of providing all the natural choices the user
expects and to make them immediately visible and recognizable. The designer needs to carefully list all
the choices a user might expect upon arriving at a Web page. The choices break down into several
categories, including process, navigation, help, and communication.
No Gratuitous Distractions. Web page hosts are quickly learning that users ignore advertising and
distractions, especially when they are on a navigational page. Even more strongly, users will avoid pages
with blinking, invasive, gratuitous attempts to get attention. The designer needs to judge every page by
its potential for distraction, and decide whether the distraction is justified.
Streamlined Processes. Business processes must be designed from the ground up to work seamlessly
on the Web. Ordering a product, tracking its delivery, paying for it, returning it, and requesting
support for it should all be part of the same overall process. Customers should not need to enter their
name more than once. Each process should know about the existence and status of the other processes. A
customer should be able to resume a session after an interruption. A customer engaged with any one of
these processes should be able to link to the other processes. The limited display surface of Web
interfaces makes this requirement more urgent because the processes usually need to be seen in isolation
from the others rather than all at once.
Reassurance. In a linear process that is hard to visualize, the status of the process should be
updated and presented from page to page. The status should be worded something like: “You have three
items in your basket. You can check out and arrange payment at any time, or complete filling your basket.
Your basket will be saved if you leave the site.”
Trust. The designer should clearly state and honor the Web site’s policies for using the
customer’s identity. Customers’ trust will build gradually, the same way they come to trust a
physical retail store. If the customer has repeated good experiences and feels that the trust has not
been abused, a good relationship can grow. Because of the multitude of choices on the Web, and the
reluctance of customers to trust a remote anonymous entity, a trust relationship that has been abused may
be impossible to recover.
Problem Resolution. A good Web user interface allows the customer to backtrack, correct previous
entries, determine status, and ultimately talk to a human being. Being denied all access to human contact
is a major mistake. It is a clear statement of not wanting to talk to a customer, and not taking
seriously the customer’s opinion. A corollary to enabling problem resolution is providing complete
information. People call about product features when they don’t get a complete description of the
product online.
Communication Hooks. A good Web user interface provides multiple links to people and sites.
Furthermore, it will provide URLs that are easy to copy and can be saved and printed on one line of an
email or document.
International Transparency. The Web is profoundly an international marketplace. A Web site with a
strong international presence should provide multiple language versions of its site to serve major
foreign markets. This feature implies support for international languages, character sets, addresses,
phone numbers, currencies, dates, and times.
Common Denominator Compatibility. Web sites must be designed for simple Web browsing software.
There must be strong support for old browsers and small display devices such as note takers or handheld
devices. One or two experiences with a site that requires a special software download or causes a
JavaScript execution error is enough to put the site on a customer’s permanent “to avoid”
list.
|
Ralph Kimball, Ph.D., co-invented the Star Workstation at Xerox and founder of Red Brick
Systems, works as an independent consultant designing large data warehouses. He is the author of The Data Warehouse Toolkit (Wiley, 1996) and the newly published The Data Warehouse Lifecycle Toolkit (Wiley, 1998). You can reach him
through his Web page at rkimball.com.
