> Those not motivated by funds could publish a zero knowledge proof instead of moving the funds.
I think at least some financial incentive is needed. It could be something like 10k satoshis, or more, just to have a non-dust anchor for the puzzle, where other people could attach more coins. For example: if you have a puzzle, where the creator cannot move the coins, without solving the puzzle, then it is still possible to increase the reward later, as long as coins are signed with proper sighashes. See:
https://mempool.space/tx/8349df0753e80cce322322f1b76789e1d0fd6693aed2f4de4e49576423081ae7In this case, the puzzle from bc1qts0jh2d2nesmketmw3thedwrx939k2tqu04gy90x9hd4049g4uhs83ltjx uses SIGHASH_ANYONECANPAY, and is grinded by the solver. And bc1q9nw84cph4dn2pw78pnaec68c82h7pxvv55cla8 is someone, who bumped the reward, by signing it with SIGHASH_NONE, so the signature is valid only in the context of the puzzle.
> Or the whole scheme can be deployed to testnet or signet not to waste (or burn?) real coins.
We also did it:
mainnet:
https://mempool.space/tx/aba3c2ae442aa20150996ee68f9aa4da83b57a4312891078be0c2e68c50b2801testnet4:
https://mempool.space/testnet4/tx/cc159432ffb7a166abeccc79800e9616a09ea9ac6937080c2ca37b38671970e5signet:
https://mempool.space/signet/tx/5aa27f3639c2218ee710fee9830924a129098dca94446661f37cae79c4e76712However, if you think, that test coins are not traded, then you are wrong. For example:
https://altquick.com/exchange/market/BitcoinSignetI think, sooner or later, all test coins will be traded, and just handled like all other altcoins. To some extent, it can be prevented, by making next chains, and dropping previous ones, like it was between testnet3 and testnet4, but still: exchanges will list new coins faster, than Bitcoin Core will release a new version, where the support for some old test network will be dropped.
Which means, that creating the puzzle will always have a cost. Because even if initially some test chain will be worth exactly zero, then it will gain non-zero value over time. And because even signet uses Proof of Work, then creating any puzzle will always require at least some computing power, even if it will be negligible, and doable on CPUs.
> plausible deniable for the publisher
It can be done, by using SIGHASH_ANYONECANPAY, which we also did in our mainnet puzzle. In this way, an unconfirmed transaction with low fees can be made, and published anywhere, using any channel. And then, other people can add more inputs, or bump the fees, by solving some low hanging fruits, and claiming the easiest puzzles. In this way, the creator has to only prepare a single input, and all outputs, and then, it can be extended by the rest of the community, and pushed on-chain by the first solver (which would also set the challenge in stone, and prove, that at least some outputs are solvable, and people know, how to claim them).