I see. Then if this is an intermediary backup, so that in the case of other backups being compromised, one could sweep their own wallet to a new wallet, it makes sense. If plates became illegal, and I need to flee, I've thought about getting a tattoo on my inner thigh with in invisible ink. But comparing the two, stenographic encryption would be a faster, cheaper alternative. What if you had some type of derivation path seed that was 4 or 6 words long, and was required to decrypt the correct information (seed) and not junk/padding data (other BIP39 words) to mislead an attacker attempting to brute-force extract all stenographically stored data?
Discussion
So if I'm understanding correctly, you are suggesting that fluff data be added to potentially detract? I love that idea actually
Yes, because even if the file is decrypted, there should be extra seed words to add more entropy(?). This is why I said about there being a small seed phrase needed to select the proper seed phrase. Even without adding extra words, there is [12!] or [24!] possible combinations, and thus numbering the seed words is also important data. In reality, you're storing a two dimensional array. One column is the seed number, and the other is the seed words. Each row has a seed word and the number the corresponds to it's order in the sequence.
TL;DR encrypted padding data to hide the real data is good. 👍