You are, of course, correct. Whilst my post is only semi-serious, I agree that if I seriously followed the line of thinking then the entire system becomes unusable pretty quickly.

As you can see, I am adapting my threat model to trust less and less structures as I go through the post. As such, it would only really be relevant in the event of a rapidly evolving adversary and probably not even then.

My threat model is nowhere near as tight as is suggested by the contents of the post. The post is more of a thought experiment to see how quickly you can “invent” threats to get to holding all data only on an offline hard drive in a cupboard somewhere. Lol

Exactly, but that just raises more issues. Which cabin? Where? How to stop people breaking into the cabin? Should it have a chimney to automatically incinerate the data device if anyone comes near? As people expect cabins to be in woods, would it be better if the cabin was outside the woods? And finally, is it easier just to trust Google with my data Lol