I used to think I could recall conversations, but how do I know they actually happened?
And of course it's not just conversations. One could easily verify his good memory by reciting old things he read (which he can easily check again) etc.
That seems trivial to verify; just record details about all conversations you have with everyone. I have been doing this since 2009. I have a 5 MB yaml file with details of every conversations (mostly tags), and I might have an above-average memory but yeah I definitely forget things .... (and sometimes I get some of the details wrong even if a conversation happened just minutes ago).
Another way to verify (to yourself only) is to writeup your memory of a conversation, PGP sign it, hash it, give the hash to a friend who was originally in the conversation, then have them do a writeup, then reveal the commitment and compare the two accounts of a conversation. The writeup would be better if it was done immediately after the actual conversation, because otherwise social groups have time to mingle and confuse everyone about what actually happened.
Another way is to do a public group writeup of a conversation after it happens (or an audio/video recording), then have everyone PGP sign the writeup (to confirm their agreement), then later pick a random conversation where this setup was performed, and write down your memory of the conversation. This doesn't work if the conversations where this setup is performed are all unique; probably has to be a few thousand different setups happen before this starts to test memory in any meaningful capacity.
Only on HN...
I hope to hell you don't live in a two-party consent state.
EDIT: Or interact with people who do. For example, California's two-party consent law applies to people calling Californians from other states.
Also, how do you know you can rely on your memories of yesterday or memories of things you've learned in school?
His inspiration was (I believe) Aristotle (if not Aristotle, another ancient Greek) who, after a church collapsed, recalled the names of every person who was present at the moment of collapse.
His filing cabinet method and Aristotle's (slash O'Brien's) memory palace methods sound very similar in that they give each memory a concrete place to exist and relate the memories to other memories around them thus securing their place within the information hierarchy.
I'm sure there is more to it than he says, and if I had to guess, I'd imagine it's a very sensory process (as sensory perceptions provide something much more analog to grasp onto than the relatively digital description a person can provide with words).