Many software already requires the almost latest hardware. That because the software is crap, no one has cared to do little optimization or use a better language/architechture etc. (yeah, yeah, no use to optimize, the customers are stupid and will buy crap nevertheless), or just not include unnecessary bloated features. There's no way running, say, bloatzilla in a usable way on a machine few years old. Or latest $office. But take a version more than few year old that does effictively the same, has most the features and half the bugs and it will be usable. (Not that I care about $office, LaTeX is much nicer to use and generates much prettier output.)
Personally, I hate serif fonts, and every piece of LaTeX I've seen generated uses the same serif font. I haven't played enough to learn to change that behavior, and I tend to not need to write such documents since I got out of school. (In school, I worked as a web geek, and tended to write reports in HTML and put 'em on my web page, so I could dive into a computer lab and reprint if I didn't have something at class time.)
I don't think people should have to buy a new computer every two years or so. Some cannot afford it, others have better things to use the money on (say, buying the CDs instead of the MP3s) or just couldn't care.
I got my mother-in-law an early PPC Mac to replace the late 0x0 Mac with the whirring drive and whining power supply, then gave her a NIC and hooked her up to my network, showing her Netscape 4.0, IMDB, Yahoo and Hotmail. One morning, when I was leaving for work, she stopped and told me that she went to Discovery.com and it told her "I laugh at your puny computer! Ha! Ha!" (I paraphrase, of course), so I got her an eMachines ~550 w/ CD-RW. She uses it for web browsing and writing grocery lists, and I have it sharing disks and printer and such and use it to burn CDs. It'll last her for 8 years (how old her PPC was in the summer of 00), I'm sure.
I have a 7GB and 15GB full of MP3s, and I still have a lot of CDs left to rip. I'm thinking of making one of my machines into a weak fileserver with 60GB disks and a 300MHz chip.
Often these same people who keep buying new hardware all the time argue how speed is progress and features are progress. Increased resource requirements on behalf of programs, for what could be done before with lesser resources is regress.
User experiences on slower hardware are better in Linux because it has been run on slower hardware and had inefficiencies pulled. Business throws more hardware at it, which would put more life into a PC, but only to a point. The computing power of my 386 laptop dwarfs the computing power of the machines they used to send up Apollo, but I never use it because it is too weak to do what I want it to do, and too heavy and unconnectable to do the other stuff I want to do. I agree, a 200MHz Pentium in an otherwise modern laptop would do most of your mobile computing applications fine. And a 486 running 3.1 and whatever Office was back then likely has the same responsiveness as a 750MHz PIII running WinME and Office2K. Andy Grove giveth and Bill Gates taketh away.
Requiring a number of huge fans, generating a lot of noise and eating a lot of power is not progress. Creating a powerfull system, that doesn't make noise, doesn't draw much energy, that doesn't have to be upgraded all the time and dumping IA32 is progress.
Going from pretty hand-drawn graphics to ugly polygons is not progress either. For real-time raytracing with geometric surfaces, real lighting and all I could, however, accept huge computing power requirements.
We can always find an excuse for a bigger chip.
[ Parent ]