Now the memory being inaccurate to start with does not make manipulating that already inaccurate memory "ok". There's no question about that. Though having said that, it's important to note that because memories are so malleable and inaccurate, we've actually been manipulating them for a long time. It's actually a pretty well researched area.
I mean, this is not really new.
It is also bad to ignore the evidence that memories are unreliable with or without memory manipulation.
We can acknowledge the weakness without using it as justification.
I doubt myself, but I still see the need to make decisions. Most people, faced with my caveats, allow that they can be wrong as well, and we decide things. But the problem cases are without any doubt - my caveats only prove their correctness. If presented with convincing evidence of their fault, they move on to the next thing and never learn that 100% confidence is overconfidence.
I don't know that my lack of trust in my memories makes me more vulnerable to such people or if they only react to facing others as immovable as themselves, with everyone else being indistinguishable speed bumps.
But wouldn't any practice of assessing one's own memories intermediately depend on memory faculties? I can quickly think of two dependencies:
1. Working memory (maybe this is tangential and not really on topic for long-term memory) 2. Semantic memories required to assess the legitimacy of a memory. E.g. if we were to shoot for "scientific assessments" like you said, then said assessment would rely on memories that allow for scientific assessment. I'm no scientist and don't know quite what such "scientific assessment" would look like in practice, but if you have an assessment framework you're deliberately working within, it seems like you'd have to at least have semantic memory/existing knowledge of that same framework.
Perhaps manipulation of episodic memory doesn't clearly pose such problems, but I expect that's neither here nor there.
I came across some ancient IRC logs just yesterday that let me look back at a situation I thought I remembered. Turns out I did not remember it.
I'm not aware of any evidence that we can believe our own memories. Memory is much like eyesight. There is a very complex process going on to give you the illusion of a clear and consistent picture that is at odds with what the messy and biological process we rely on can actually achieve.
The basic problem is that people don't by and large scientifically assess their memories for accuracy, so they are reliant on other people alerting them to problems. The feedback process there doesn't encourage correctness, it encourages conformance.