Fraser Speirs posted this last month but I’m not exactly on the edge these days:

"Imagine this scenario: Overnight someone sneaked into my office and upgraded an application on my computer. An application I had been running happily for months, and one that worked well and served my needs.
Obviously, nobody asked me if I wanted the upgrade. What’s more, the phantom upgrader also didn’t check that the new version was compatible with my Mac. Okay, maybe they did test but decided that the wrinkles weren’t all that bad and that I could probably live with them until they got around to sneaking back in again and doing another drive-by upgrade.
Sounds absurd, doesn’t it? Absurd, yet that’s exactly what happened to me a couple of weeks ago. I was preparing for a trip to San Francisco and mapping all kinds of things – my hotel, the conference venue, my friends’ houses and the place I’m going to show off my awesome born-in-the-UK bowling skills. Then Google decided to roll out a new version of Google Maps and it didn’t work cleanly with Safari.
Here I am, in the middle of an important organisational project and the single most important application for that project was just given a disruptive upgrade. Did anyone warn me that I should expect and plan for disruption? No. Could I pull out the installer CD and go back to an old, known-good version of Google Maps? No. Could I long-term refuse to upgrade on the basis that the current version met my needs and I prized stability over features? Not a chance."

Many people will eventually realize that mainframe computing, even renamed web 2.0 and wrapped around in in the latest verbiage, is not exactly a liberating model. There’s a long way before web-centric architectures provide the kind of control you take for granted on a PC.