Of course. The same did the guy in testnet3 that mined those transactions with nine leading zero bits. So now we know that people can mine transactions, but still, the whole solution is unknown.

Of course. Keep your secrets, that's the whole fun to guess all of that. Also I know that you wrote exactly that sentence before. So it is like OP_CODESEPARATOR: the secret part of the script is already hidden, only the remaining part is visible.

But I still wonder, if it is possible to make it easier in the future, by revealing some hints in decentralized way (also, some sighashes can be used to make it an incentive, like "I will reveal you my public key, only if you put N coins to this challenge", it is possible to make it so that coins will move only when both parties will meet those conditions). For example, spending "OP_HASH160 5ab6db66fb0dd08ecb238f07141ffea56189552d OP_EQUALVERIFY OP_CODESEPARATOR <newPublicKey> OP_CHECKSIG" could reveal the public key for the puzzle in a trustless way. And later, any 2-of-2 multisig could reveal that the private key for the puzzle is reachable. Because now, you know, only some random hash is visible. It could be a trap address. A public key could be unreachable or invalid, or a private key could be provably unknown. Or maybe everything was created as a result of SIGHASH_SINGLE bug, or a tricky signature: https://bitcointalk.org/index.php?topic=5373858.0. Nobody knows, and nobody can trust any future hints, that is clear. But even if there will be some disinformation, it can be still used to make more challenges.

For example, I can tell you it is not based on sethdseed, as far as I checked. But it seems like another nice challenge, to start from sethdseed true "KwDiBf89QgGbjEhKnhXJuH7LrciVrZi3qYjgd9M7rFU73sVHnoWn", then do some tricks, and reach some interesting brainwallet. For example, here it will be always bc1qvl6xr6th5v7qafuclkkvv67w803hw5zk5g60xu, and then, it is possible to deterministically do many tricks.