Interesting remarks on legacy tech (Posting: Hal Kaplan)

From this thread:

When I was a newbie, I worked for a company that kept two RCA Spectra 70
systems around way past their useful life because the systems hosted two
applications: Billing and Payroll. The programs were written in COBOL.
But it wasn't just COBOL or Spectra 70s. It was the disk drives, and
the printers. This company made several deals with the devil to keep
these machines going. It just could not bring itself to cutting the
cord. The vendor (Sperry) FEs would spend hours each week scouring the
nation for spare parts. We had printer ribbons up the kazoo stockpiled
because they were generally unavailable. Oh yes, they kept the
programmers around too (sweet job, I guess). And they postponed the
demolition of the old corporate headquarters to avoid having to move the
Spectras.

So I guess times are changing, but why do they have to change?

If you have an application running on a computer configuration today and
that running application is producing a meaningful product or service
for your company, why do you need to switch? When did the "upgrade or
lose support" gene become part of our collective DNA? Support is an
issue for immature products. When was the last time you actually got
support from M$ that was of any value? Probably shortly after you
upgraded (or updated) your system and the recent change was not yet
ready for distro.

The real problem here is not M$ but Intel. M$ stuff will only work as
long as Intel (or AMD, I suppose) produces a product that will bootstrap
the way it does today. Yes, one minor circuit change and everything,
including Linux, will fail. And that failure will be a lot harder to
overcome than if M$ changed one line of code.