But was your son doing a 20-year study? Was he using real-time laboratory hardware? If he wasn't doing these, using new equipment is fine.
Any study that relies on the same computer hardware/software being available continuously for 20 years is likely doomed to fail before it even starts. Creating an experiment/study that relies on hardware that depends on specific bus timing is extremely bad design. I'm having trouble conceiving of an experiment that would have to conform to the restrictions you list.
If the computer is replaced and the old lab equipment will not work properly with it, the entire study so far is turned into to garbage.
That's why you determine that the software will run on the new equipment before you decommission the old equipment.
Reverend Jim seems to think that all you have to do is plug the new computer in, install the software, and it works.
That's not even close to what I was saying so please don't put words in my mouth. I spent 15 years developing/maintaining software on an AGC/SCADA mainframe that was tied to dozens of RTUs (remote terminal units) at many generating/distribution stations around the province. We knew that eventually we would be unable to keep the system running due to the lack of spare parts. We eventually started a replacement project that went online in November of 1998. The new system went through a strict commissioning procedure before the old system was turned off. Even after the switchover, we kept the old system functional in the event of a failure. Hardly plug and play. The system is used to control all of the generation and distribution of power in the province so it is magnitudes more complex than the systems running at missile silos. Incidentally, the new system used the same communication protocol as the old system to talk to othe RTUs so there was no incomatibility. I will grant you that certain equipment failures at a silo can have more devastating consequences. In any case, I do have some experience with systems that must remain functional over long periods.
I should also mention that I have developed software that has run on several generations of hardware. In a few cases only minor software changes were required.
And if the old computer really has 8" disks, it doesn't have any of the microprocessors that are in use today. It probably has an Intel 8080, a Z80, a Mostek 6502, or a Motorla 6800 or 6809. Any software that runs on these will NOT run on any new computer.
So give me an alternative. You can't keep ancient hardware running forever. Eventually it has to be replaced. Everything wears out.
What I want is the ability to buy BOTH kinds of computers.
Are you saying that you want the manufacturers to continue making the same computers they were making back in the 70s and 80s? What company could stay in business with that strategy? Do you see Ford still making Model Ts?