I think a lot of people hear about the method of loci at a young age but then ignore it. I certainly did when I was a kid. Probably every 5-10 years or so I’d pick up one of these memory books and see if anything new had been invented, and I was generally unimpressed because it seems like it was mostly useful for memorizing random things and not information that actually has meaning. So this would have been in the 70s and 80s. Books like Sheila Ostrander’s “SuperLearning” and other stuff like that were not attractive to me because there was not context I could glean from people being really successful with them. Plus, without motivation I found I could practice those techniques for maybe 30 minutes and then reject them as being too ponderous. I think one book I read even had all the pegs mapped out, which is silly because may were not meaningful to me or easy to remember at all.
Now if you are a synesthete, maybe the first time you hear about the method of loci your initial success is greater and that encourages practice. So it could be that S fell into this practice at at early age and didn’t think about about it.
One thing that impressed me about S vs. professional mnenomists is that Luria tested S on random syllables without warning and S was still successful at those. A trained mnemonist might have to practice a system for that ahead of time. Now it’s possible that S was clever and somehow found out that would be one of the tests so he could prepare something, but to be effective at random information for which you have no previous encoding and no chance to practice seems fairly impressive to me. Imagine if you showed up at a competition expecting to memorize decimal digits and then they changed the rules and announced, “okay, we’re going to memorize emoticons instead.” How well would you do?