Sean Nelson
Forum Pro
I agree that problems of this sort are few and far between, but they do happen. I started having verification problems on my backups that I initially blamed on the disk, but when they kept happening with different drives I was forced to troubleshoot further and eventually discovered that I had an intermittent memory problem. The only thing that caught it in normal use was the end-to-end verification of my backups. The POST memory test at startup missed it, it wasn't until I resorted to running lengthy memory diagnostics that I was able to pin it down.You are worrying about a zero-risk problem in practice, and you can't verify 95% of the file movements that are happening all the time and that you are not aware of.In theory, explorer is fine. In practice, stuff happens. I prefer to trust, but verify.
Read-after-write is useless in this kind of situation because the error occurs before the data is sent to the disk. Alas, memory is the single point of failure where zero error checking is done in most consumer systems. And I run everything to spec, those folks who play around with overclocking are more at risk.
As a result of this experience I now only buy systems with ECC memory. That will hopefully prevent this kind of problem from going undetected again. but Murphy lurks under the covers and for the irreplaceable data I've collected over a lifetime a policy of "trust but verify" keeps me happy. It's not as if I have to do it by hand, I have this thing called a "computer" that does it for me while I do other work.
Each person can decide for themselves what precautions to take based on how much they value their data, but they should base those decisions on an understanding of what kinds of things can go wrong.