In the late 1960s, Laurence Peter created the "Peter Principle", postulating that managers are promoted to their level of incompetence, causing organizations to falter. As I reflect on the software tools I'm using in 2005, I've concluded that there is a software corollary to the Peter Principle -- software evolves to the point that it's unusable.
I'm writing this column in Notepad. Why? Have you tried writing an outline, end notes or an indented, bulleted list using the latest wordprocessing software? Wizards and autoformatting tools try to anticipate what you're typing and in the process irreversibly scramble your work.
Our modern operating systems contain vast numbers of CPU-consuming add-ons: a wagging dog that searches for your files, invisible background processes that constantly download patches and user-interface gizmos such as thumbnail previews of your multimedia. With all this increased complexity comes a lack of reliability, perpetual security holes and poor performance. Boot times are long, lockups are frequent, and viruses are epidemic.
Although my computer today is 100 times more powerful than the one I had in the late 90s, my current environment has less speed, lower productivity and higher cost of ownership than my Pentium running Windows 98 Second Edition and Microsoft Office 97.
In my view, it's time to rethink what the industry is producing with thick-client software, bloated with a spiralling number of esoteric features.
This is not about Microsoft vs open-source. It's about creating highly reliable, usable tools that run anywhere, anytime. It's about reining in the sales and marketing departments of software vendors whose revenue-growth targets propel them to offer feature-filled upgrades more often than the customer base desires.
At dinner with Microsoft executives last year, I asked them to consider a lighter, cheaper, highly reliable version of Microsoft Office. They responded that surveys indicate that corporate customers use 90 percent of Office's features but I suspect that a closer look at this data would illustrate that only a very few power users need advanced features.
In 2006, let's break the cycle of creating more complex, less reliable, less usable software and agree that less is more. I encourage the software industry to take a lesson from Gmail and other successful thin, good-enough applications. Do we need Longhorn/Vista and a new 3D graphics engine-driven user interface with so many lines of code that it will be a challenge for even the most brilliant programmers to maintain?
We need the Toyota Prius of software -- "Google Office with Ajax for Linux" or "Microsoft Office Lite" -- and not the Hummer.
John Halamka is CIO at CareGroup Health System and CIO of the Harvard Clinical Research Institute