My question is, do people trust write-blocking devices to do what they say they do, or do people always test them out?
What I'm interested in finding out are what tests are deemed to be reasonable when testing the write-blocking capability?
Yes, YOU should always test your gear, then you are fire-proof in Court.
How to do it? Why not take an MD5 of a drive, put on the write blocker, use the drive for both reading and writing, then do an second MD5. If the two MD5s are the same no alteration of the drive has taken place.
That's one way of testing it. Would you carry out this just the once or before the start of every case?
I don't think it's reasonable to validate all your software/hardware before every case. Just do it when a new major version is released in the case of software, and perhaps from time to time when talking about hardware.
No, that is absolutely right, you only need to test when something changes e.g. new hardware or new release of the software. The whole principle of what we do is one of repeatability. If you do the same thing two or more times you will always get the same result.
The test that I suggested above was, of course, not rigorous but more pragmatic. The problem with testing is that the more rigorous it becomes the more the tester has to understand the inner workings of the system. If I produced a program that exercised the disk interface completely (what does completely mean?) and you tested the write blocker with that program, you would not really be any further forward unless you saw and fully understood the inner workings of my program until many people had tried it and found it to work correctly.
We therefore rely on each other and our own pragmatic tests. We are content that a tool works if enough of agree that it does and it appears to us, on our own testing, that it works.