YOU ARE HERE: LAT HomeCollections


Linux May Be the Friendliest OS of Them All

August 21, 2000|CHARLES PILLER

I've recently been feeling cautiously optimistic that the personal computer might eventually be freed from its mind-boggling complexity and ponderous inefficiency.

Things were once far worse, of course. Before the Macintosh, the user was treated as an afterthought. Apple turned this around with the first mass-produced computer that began with the user interface and built the hardware and software around it--finally making computing accessible to almost anyone.

Unfortunately, neither Apple nor anyone else has done much to fundamentally improve on that 16-year-old Big Idea.

The original Mac brain trust left Apple years ago, but three of its brightest lights, Andy Hertzfeld, Mike Boich and Bud Trimble, have joined forces again at Eazel (, a Mountain View, Calif., start-up that previewed a graphical interface for the Linux operating system last week; it will be released later this year.

Before I explain why Eazel and other Linux developers could make computers easier, consider some basic problems of interface design, and a radical alternative outlined in a new book ("The Humane Interface," Addison-Wesley, 2000) by Jef Raskin, original leader of the Macintosh development team at Apple.

For the Record
Los Angeles Times Tuesday August 22, 2000 Home Edition Business Part C Page 3 Financial Desk 1 inches; 16 words Type of Material: Correction
Eazel executive--The name of Eazel Inc. executive Bud Tribble was misspelled in Monday's Innovation column.

Raskin, a cranky contrarian, finds little to admire in the Mac OS or Windows, or in software applications such as word processors and Web browsers.

To understand why, consider what it would be like if autos were built like computers. To appeal to all drivers, each car would come with both manual and automatic transmissions--you could use either or both at once. Shift into third gear when you should be in second, and a warning message would obscure the speedometer until you click "OK" on a touch screen in the dash. Then the display would offer unsolicited tips on how to customize your gear shift to hit second gear reliably.

Of course, car makers (and liability lawyers) understand that controls should be unobtrusive and forgiving so that drivers can concentrate on the road.

Yet computer users accept a multitude of similar design follies, such as:

* Useless dialogue boxes that require a response but offer no remedy for the problem detected.

* Distracting "progress" indicators that indicate . . . nothing. When copying files in Windows, for example, the progress bar bears no relationship to how much time is required to complete the operation.

* Confusing redundancy, such as five different ways to launch a single software application.

* Rapid obsolescence, which makes Web sites unusable without "plug-ins" that can take minutes to download and often conflict with existing applications.

* Differing controls for every piece of software--mediated through an equally complex operating system--coerce you into being a computer operator, distracting you from real work or play.

Pin part of the blame for this mess on "backward compatibility." Each time a software maker adds a new option, it keeps the old ones, no matter how inferior, out of fear that nostalgic customers would rebel. Add commercial imperatives--software is sold for "new" features--and you get a muddle of complexity.

Windows and Microsoft Office together contain more than 1,000 user-controlled settings, according to Raskin. Today's software has become a tyrant, taxing your time the way the IRS taxes your paycheck.

The monotonous interface--such as the fixed controls of a car--is the easiest and most efficient, Raskin argues. Some of his solutions seem self evident, perhaps explaining why they've eluded the industry so far:

* Hide the plumbing: Settings, trouble shooting and most other futzing should be automated and hidden from the user.

* Don't force users to be experts: The customization craze with software programs and Web sites expects every user to be an interface designer. With experts faring so poorly, why expect the rest of us to do better?

* Be forgiving: Human beings make mistakes. Every PC should have a lasting repository for all deletions, and a universal "undo/redo" command should be hard-wired into every keyboard.

* Kill the icon craze: Many of those cute little symbols are incomprehensible, and with hundreds on your screen even good ones lose their power.

So why might Linux--a system known for extreme geekiness--do better?

Unlike Windows or the Mac, Linux is "open source," created and tested by thousands of volunteers. It invites experimentation that is impossible for a single company. Linux is also free, offering an escape from commercial feature-itis. And because Linux is at the genesis of its interface-design period, it need not be a slave to backward compatibility.

That combination offers hope that Linux will take genuinely new directions. It had better. A mere copy of the Mac or Windows interface would ensure its failure; to win over typical users, Linux has to be far better.

Eazel's Nautilus--the graphical interface for Gnome, ( a project that is building user-friendly, Linux-based software--shows some promise along these lines.

Los Angeles Times Articles