Spectacular Failures: Firefox 4 and LibreOffice

Yogi Berra, the baseball player and manager, once said, "We're lost, but we're making good time." We all know people who are so obsessed with speed that they actually slow down. People who don't take seconds to document their work. People that don't take minutes to run some tests. People who don't take an hour to plan out some skeletal data designs. A few months ago I heard of a manager who fired a team member for becoming sick during a time-sensitive project. Computer professionals are always concerned about getting work done fast During a major site outage or a 48 hour deadline for a proposal, there's no time to worry about the consequences of our actions. But projects slow down as they get larger and more complex, and for any project that takes more than a few days, documentation and strategic design is done to speed the final deadline, as as Fred Brooks points out in "The Design of Design". Often times developers who brag about their coding speed have designs that cause the longest delays.

We had a couple of recent examples of major open source projects that may have crossed the line from speed into splat.

First, Firefox 4 was released. The flagship browser of Linux and the tool of choice for most web development professionals, this new Firefox was described as having "super speed" and an "improved interface" (Firefox Features Page). What Mozilla delivered was a little different than advertised:

• The new look was lifted directly from the Opera web browser's clunky user interface. The new Firefox hid important features, and made page navigation difficult (Firefox 4 Finally Released, and it Looks Like Opera, HowToGeek).
• A browser that locked up many Windows computers. Announcing that they were following an aggressive release schedule, the Firefox developers released Firefox without properly testing their graphics acceleration routines, causing some fonts not to render properly and would lock-up some Windows computers entirely.
• Linux users got the shaft. Many of Firefox 4's improvements came at a price that they were Windows-only and no attempt would be made to guarantee stability and support under Linux. Linux users would see no "super speed" for months, if ever. (Why is Firefox slower on Linux than Windows?, Internet.com).

Many of the web developers I know have been switching to Google Chrome or other browsers, not because they don't like Firefox, but because it's buggy and is hard to use.

Then there's LibreOffice. This project began in the late 1980's as StarOffice (Wikipedia), a suite of office tools like a spreadsheet and a word processor. It was first offered for free in 1998. Because it supported both Linux and Sun Solaris operating systems, Sun acquired StarOffice in 1999, renamed it OpenOffice and released it as open source. For 10 years, OpenOffice was the premiere Linux office suite.

Then, in 2009, Oracle bought Sun Microsystems. Besides OpenOffice, Oracle also acquired other open source friendly products like Java and MySQL. This created a wave of paranoia: what if Oracle decided to close source all of these products? It would be a terrible blow to Linux.

Despite repeated assurances from Oracle that they had no intentions of restricting the licenses on these products, 20 of the OpenOffice developers went into irrational reactive mode, forking development (OpenOffice.org offshoot LibreOffice debuts, CNet). OpenOffice became LibreOffice. Another team turned MySQL into Drizzle. Drizzle was a fizzle: no company would switch from MySQL unless there was a reason, but most Linux distros forced their user base to LibreOffice, whether the users wanted to or not. Users were confused, such as author Piers Anthony, who wondered in his April newsletter which product he was supposed to use.

Within 3 months of LibreOffice's release, Oracle announced that it would no longer support OpenOffice (Oracle orphans OpenOffice offering, ZDNet UK). LibreOffice supporters saw a smoking gun: Oracle was trying to kill OpenOffice. But others insiders pointed out that, with most development and support switched over to LibreOffice, Oracle was left with an orphaned product. LibreOffice's success may have brought OpenOffice's downfall, creating the very fate OpenOffice supporters had been afraid off.

Two spectacular failures, victims of their own success. One successfully released too fast and created an under-tested product that alienated its users. The other forked too early, confusing users and possibly killing its own parent without cause. There ought to be lessons to learn here about thinking before coding, but I'm concerned that these teams are too busy making good time to realize that they are lost.

May 25, 2011