Would you please point me to the txids?
with nine leading zero bits
Sorry, bytes of course. There are more transactions, but you can start with them:
https://mempool.space/testnet/tx/0000000009e6f27e34e9d5e17df63758375195eb76b46daf0327b4aba151bfb4https://mempool.space/testnet/tx/0000000005771270ab7be0d9dc5358048aad0e003c6482819438fd9a61b80841it may lead the way to the solution, or it be a red herring
In my opinion it is just a nice trick, but I could be wrong. Also note that it is possible to use sequence number or locktime to make it easier. For example, it is possible to set sequence number to 0xffffffff, then locktime is ignored, and can be used as a nonce, just like in a real mining, only the total size is different. But even if someone has an ASIC miner, then it is possible to choose the right transaction size, to optimize it for a specific hardware.
Someday, this “mining” of txids may make for an interesting thread in the development forum.
When it comes to transaction mining, it is just a nice trick, but nothing else is behind that. The better way would be to mine on top of the latest block, then by using Merged Mining, it could be possible to reuse Bitcoin Proof of Work, and make the whole network stronger. And then, it could be a part of decentralized mining, where people could receive N times less coins for mining N times easier block headers. In this way, CPU mining could be used to get a single satoshis or millisatoshis discount, when using Lightning Network.
Was that the first time anyone had ever done this?
Of course not. See this testnet3 transaction:
https://mempool.space/testnet/tx/16fb1c624b3658c78630c39f1c33e8f3dbdb71373f33eb22c6abe4405c0815bcYou can try to grab some zero satoshi test coins by re-mining satoshi's blocks. And there are more such transactions.