>I think you've made some excellent points, which inspires me to add my 2 cents. There's no sense in asking whether there will be "defect-free software" without also asking "for how long", but this is a rather theoretical debate.

Being somewhat nostalgic, I still keep some very old games around. Surprisingly, the survival age and the halftime of decay spread equally, regardless of the time and the current OS and HW version they were written for. Some have not survived DirectX, some have become too fast on 386s or even 286s(being timed for XTs), some were using dirty tricks and clashed with EMM, QEMM, smartdrv or NT (whichever was rigorously guarding its memory quarters)... and there's the champion, the sole survivor from 1989 or 88, still working without a glitch: bl.exe, the 3D Tetris, written by two Polish guys (Ustaszewsky and another one). It worked through all versions of Dos and Windows, including most of the W2000 installations, had usable picture on any MCGA, bw Hercules, EGA or VGA card. It worked same speed from XT to P3's (except some EGA cards made it somewhat slower). Its only bug (or a feature) is that it doesn't allow spaces or any punctuation in the names in its hall of fame.

If there's any morale to this story - if you're looking for bug-free software, look at the games. That's where the authors set the rules, and they are in charge of defining what is a bug and what a feature.

For us, mere mortals (may my neighbor forgive the pun), it'll always be end users who will dictate what will the software do, and... you asked for it, you got it. New features will be introduced, even if the feature which will introduce new bugs is "should be able to run on the next version of Windows". Hard is the life of us artists.