Today Mircea Popescu offers a proposal which he is mulling as a necessary condition for any fundamental change to the Bitcoin protocol to receive his approval. The proposal would strengthen the bond between Bitcoin miners and nodes. Popescu is pledging to sink any hard fork proposal that does not include these changes, but if these changes are to be adopted he is willing to negotiate changes to the Bitcoin protocol including a maximum blocksize increase.
 |
Qntra - Herald Of The Most Serene Republic |
This idea is somewhat interesting, but the goal sure is damn vague.
It certainly makes block validation far more burdensome. Every node on the network (not just the miners) has to do this O(chainsize) I/O-intensive operation every time a new block is found.
> that the miner-node division
This is pretty vague. Are you trying to stop dumb Chinese fuckwads from SPV-mining? Sure this will achieve this. Sort of like a nuke will get the grout out from between the tiles in your bathroom. Quite a major change simply to stop irritating behavior by two mining pools.
> although it does have the marked advantage of making ASICs altogether impractical for mining
This is a reasonably well-defined goal, although it isn't mentioned until a footnote. I'm also not sure the proposed scheme achieves this goal. There's a great likelihood that it merely requires different ASICs. ASIC-resistant PoW is still an open problem and the field is littered with the corpses of failed schemes. None of the as-yet-debunked ones have stood up under the withering heat of a lucratively-mineable chain for long (or at all).
Also remember the nonce field isn't the only place you can stick nonce bits… it's merely the cheapest place. If you're willing to do twice as many double-SHA operations you can just mine every block with nonce=0x00000000 and use the coinbase or some throwaway transaction as a source of variation (hence, parallelism). So whatever you're trying to force miners to do here, that's the cap on how much it can cost them. If whatever you're asking for is more expensive than simply halving their hashpower (like, for example, baking new ASICs), they'll simply ignore you and put the nonce elsewhere and "buy" (or produce themselves) only one of these known-nonce-blobs every ten minutes.
gbanGASIC
s/as-yet-debunked/as-yet-undebunked/
MP requests you please offer your criticisms and insights on the Trilema post as well.