I am very much a blockchain skeptic. Basically, most of the benefits are illusory and the risks are considerable. It doesn’t replace the need for governance. It doesn’t decentralize nearly as much as it promises to. And, near as I can tell, none of its applications truly need its security properties.
This is a brief paragraph in the context of an Ask Me Anything. He promises a longer essay forthcoming.
I’m with Schneier on blockchain. My employer is currently trying to get into that space. They’re very excited about this new startup that makes blockchain verification orders-of-magnitude faster.
How do they accomplish this? By re-centralizing the network and having 20 master nodes which do the verification!
One stop check on liquidity, account balances. Just check the latest copy of blockchain.
Merkle tree is likely one special case of the generic multi-access consistency systems, and all the bank computers have some kind of protection against the race condition.
Let us take a smaller example, a group of banks decide to release digital cash, let it float while holding the corresponding official deposit account in freeze.
So, a group of stock traders decide to dispense with the bank accounts all together, they just trade digits for stock using their telephone. The banker let them use the digits for a hundred trades, it must have a limit in time and count. After the digits passes 100 times then they have to restart. with some fee.money
Who check the digits for double sending? In this case use a more finite consistency check. Use a tree, of some sort, and delete a digital cash trace when the digits are returned to the bank, thus shortening the tree. The typical trader need only check a local tree with no more than 100 steps in any account. In this case we have a time varying tree guaranteed with no path longer than 100. Just another, better way to do multi-access SQL And SQL does something similar when it prititizes requests, it builds software tree.