The words "update" and "upgrade" are dirty words to me.
I was the engineer for a set of 9 scientific labs at a university. Then the IT department decided to convert every computer on campus from DOS to Windows. They came in at night and changed out all of the computers without telling anyone.
The next day, nothing in the labs worked.
- Since the data collection cards would not fit the slots in the new computers, they just left them next to the new computers.
- The data collection cards were $2000 each.
- The company that made the data collection cards had no Windows-compatible versions (and never really solved the problem of making cards that work right under Windows).
- The data collection drivers were designed for DOS. They wouldn't work under Windows.
- The cards for DOS worked to 1 millisecond data rates. The company could not get any cards to work in Windows for any rate faster than the 55 ms (about 1/18 second) multitasking rate.
- None of the custom-written (by me) software to control various experiments could be used. Our software was based on reading the data ports once every millisecond. That way, we would not miss any changes and would have them at the exact time they happened and could instantly issue commands to adjust the process.
- Windows is based on time-stamp data collections. They say that it is "more efficient" to collect data in bursts with time stamps to indicate when they really happened. But:
- - How to you "tell" the chemical process that the reagent should have really been added 600 milliseconds earlier?
- - How do you tell the organism that it should have been stimulated 200 ms earlier?
- No other computer operating systems were available that do not multitask.
Their "upgrade" totally destroyed the lab systems. Science stopped until we found a solution. Their office-based brains could not understand why we could not use computers designed for office data processing for science.
We finally talked them into giving us back all of our DOS systems. But they said that they would no longer maintain the computers. So I kept them going for ten more years. But as they failed one by one, I could not get replacements that worked with the data cards and DOS.
The result of the upgrade craze:
It is impossible to follow the ceteris paribus rules for scientific studies for any scientific study using computer controlled processes and data collection for any study longer than 5 years.