Ran across this article on /. a while back: http://www.eweek.com/article2/0,1895,1923402,00.asp
The author makes the somewhat controversial statement that what ends up on users desktops isn't what's best, but what came there to begin with. I tend to agree, based on my experiences with "end users".
The average end user still doesn't understand exactly what a browser is, or why they would want to try a different one. To many home users especially, Internet Explorer is the internet. They do not mentally separate the program from the activity. (The tool from the task.) This is normal for first time PC users. You want to do task XYZ, well you click here on program ABC. Mentally ABC == XYZ. It's only over time and with experience that they may learn they can do the same task XYZ with program DEF as well! For some people, this is like the dawning of the day, and they begin exploring what alternate programs they can use for the other tasks they do on a regular basis. They soon learn that different programs do the same task better, or just differently, have different features, etc., and find the programs that best fit how they want to perform their tasks.
For others, this situation presents overwhelming choices, and is a very bad thing. These are the people who get frustrated with the computer when they have to choose what program to use. It's like ordering coffee at Starbucks. (Or any food from a replicator on Star Trek.) Sometimes you just want coffee, without having to add 15 modifiers to identify a specific drink. They want the computer to just do certain tasks, and do them well, without asking technical questions about how to do them. Layers of obscurity into how things work are not always a bad thing, at least, not if they can be gotten around easily if the user so desires. After all, this same principle lets you drive a car without understanding how a fuel-injection system works. Do you really care what brand spark plugs are in your car, or know if they are gapped properly? Some people do, most don't. If you had to know to drive a car, would you be finding another way to get to work tomorrow?
Usually, single task devices seem to come first, and are extended into multi-function devices, like a cell phone that can check my e-mail, browse the web, play music and movies, and let me instant message people, plus, oh yeah, call someone. However, the argument can easily be made that this is often at the expense of whatever the primary function of the device was originally, and almost always at a price of a steeper learning curve and more complex user interfaces.
However, some people just want a device that does one thing, and does it very well. Do you want your checkbook to play music and have net access built in? Do you really need a TV built in to your fridge? It is my opinion that at some point the idea of computers as specific machines we sit down at to do things will have to fade away, and be replaced with a type of distributed processing. The processing will all be done in the background out of sight, and we will simply perform our activities where they are most natural for us to do so. If I sit down at my desk at home, I can access my bank accounts and pay my bills. I can also do it from the couch or outside if I want to, but it will be what the system assumes I want to do when I sit down at my desk, based on my normal routine. The same system will display recipes for me in the kitchen, if I want, and will have my music follow me as I move around the house, but not into the kid's rooms when I peek in to pull their covers up after they are asleep.
If a system becomes this pervasive and integrated, most people won't want to know what recipe program they are running, or if their lawn maintenance software version is compatible with their new robotic lawn mower they just bought. Yes, some people will know, care, and love every minute detail. They will have custom interfaces for everything, and their houses will literally respond to their every whim. I will probably be one of these people. However, this will be the exception, because everyone else will just want it to work, and won't care if they are using Blinds 3.0 from Windows Corporation, OpenMyBlinds 4.27 from AOL/Time Warner, or GBlinds 2.7 (Beta) from Google World Domination, Inc.