4 ms·

> BFT-style consensus such as Paxos and Raft Raft cannot deal with byzantine failures, so it's not a BFT protocol. Neither is the original Paxos, unless extend

> BFT-style consensus such as Paxos and Raft Raft cannot deal with byzantine failures, so it's not a BFT protocol. Neither is the original Paxos, unless extended to Byzantine Paxos. Both are CFT by default. > So, we would not expect distributed state machine replication to get better by requiring permissionlessness. Right, but the banking system isn't a 600LoC Raft implementation. By your own admission, the bulk of inefficiencies come from the other 99% of stuff banks do. So the question is: how much regulatory burden can be reduced if we were to switch to a crypto-based infrastructure. > just so you can circumvent the law. I can pay taxes in crypto to my local government. We also have crypto ATMs in several places here. Blockchain companies are working very closely with financial regulators. Not sure what kind of laws I'm supposedly circumventing here.

FabHK 2y ago
By BFT-style I meant employing rounds of voting rather than longest chain, thus being always consistent and eventually available (unlike LCR, which is always available and eventually consistent), but fair enough. Anyways, in the permissioned setting this was all solved in the 90s iirc. Then, so, just moving stuff around in a robust and even distributed ledger is, we can all agree, while not trivial, a solved problem. So then one can impose further constraints. Either, 1), comply with the law and rules and regulation, and sure, that will make it less efficient. But that is the 99% of the work banks do that crypto doesn’t do, and they are not being done for fun, but for good reasons. Or 2), permissionlessness, but what for? So you can avoid 1), because otherwise you wouldn’t need to incur the tremendous costs of 2). I don’t think we should adapt the AirBnB/Uber approach (how much of the regulatory cost of fire regulations/working time restrictions/insurance cost can we avoid) to money transmission. It is obviously trivially possible to do legitimate transactions with crypto, just as you can cut an onion with a machine gun. But that’s not the raison d’être of a machine gun. And the raison d’être of crypto is to escape regulation, as Satoshi basically said in the original emails, and as is abundantly clear not only from looking at the last 15 years of history, but also from the willingness to incur the massive overheads of permissionlessness. ETA: appreciate the discussion, btw.

FabHK 2y ago

By BFT-style I meant employing rounds of voting rather than longest chain, thus being always consistent and eventually available (unlike LCR, which is always available and eventually consistent), but fair enough. Anyways, in the permissioned setting this was all solved in the 90s iirc. Then, so, just moving stuff around in a robust and even distributed ledger is, we can all agree, while not trivial, a solved problem. So then one can impose further constraints. Either, 1), comply with the law and rules and regulation, and sure, that will make it less efficient. But that is the 99% of the work banks do that crypto doesn’t do, and they are not being done for fun, but for good reasons. Or 2), permissionlessness, but what for? So you can avoid 1), because otherwise you wouldn’t need to incur the tremendous costs of 2). I don’t think we should adapt the AirBnB/Uber approach (how much of the regulatory cost of fire regulations/working time restrictions/insurance cost can we avoid) to money transmission. It is obviously trivially possible to do legitimate transactions with crypto, just as you can cut an onion with a machine gun. But that’s not the raison d’être of a machine gun. And the raison d’être of crypto is to escape regulation, as Satoshi basically said in the original emails, and as is abundantly clear not only from looking at the last 15 years of history, but also from the willingness to incur the massive overheads of permissionlessness. ETA: appreciate the discussion, btw.