3 ms·
I remember about 2017 or 2018 giving a talk about Ethereum and asking how many people in the audience of maybe 400 had heard of the CAP theorem. One hand went
by leashless 3y ago
I remember about 2017 or 2018 giving a talk about Ethereum and asking how many people in the audience of maybe 400 had heard of the CAP theorem.
One hand went up in the back. That's when I knew our culture was doomed.
CAP is (for the record) why the blockchain is and is not "just a database" -- it's a database which takes a different position on CAP and as a result can be used for different things to other databases.
- tomhallett 3y agoWould love to learn more. Guess: Is the “different position” that consistency can be achieved when a fault occurs by having the longest chain, instead of a leader/voting mechanism?
- leashless 3y ago"Eventual consistency" 1) there is a network 2) on each block, a leader is selected to make the next block on the basis of a lottery -- each lottery ticket is a unit of computing done ("proof of work") 3) if the network splits, each subnet has a different leader and a different sequence of blocks from that point onwards 4) when the network comes back together, a new "agreed history" is formed from the last shared block, selecting the block with the most "work" done and resulting in the transactions which were done and confirmed in the smaller subnet being thrown out and history rewritten as if they had never existed This is the genius of Satoshi -- he figured out that something this trashy was good enough to get something done. Nobody in an academic setting would ever have considered that as a worthwhile solution I don't think. It was madness. But it worked. It's the same kind of genius as Ethernet just rebroadcasting packets until something works rather than doing Token Ring type negotiation about who can talk next. Worse really is better at times.