I don't think it confirms that the memory was "real" - merely that it was useful.

Which is the same criterion that is met by social confirmation, is it not?

It seems like what your argument really boils down to is the notion that all words in general (of which "memory" is one) are inherently only socially useful, since they are how we attempt to communicate our thoughts to other people. But that's not really a fact about "memory", so much as about words in general.

Although with the invention of writing, words can be useful in non-social contexts too, since we can write notes to ourselves that may be useful later.