The five consensus algorithms #4: Voting-based by Dr. Leemon Baird

one type of distributed ledger technology is voting based DLTS pure voting based DL t's these go back decades these have been in the literature for a very long time and they had incredibly strong math proofs for them they are Byzantine fault tolerant many of them can be asynchronous Byzantine fault tolerant which is the good kind of Byzantine the strongest kind of Byzantine they had incredibly strong proofs about them they're sort of complicated but it's under easy to understand how they work when you think about them and you go through the math they have these beautiful math proofs and they have an enormous literature going back decades people are still publishing PhD theses and doing ongoing research on how to make them a little bit better each year but nobody ever deploys them everybody deploys hybrid systems that do voting based plus leaders or voting based plus something else or they do virtual voting which is what hash graph does but nobody does pure voting space systems because a pure voting based system is going to be horribly inefficient the problem is that if I have a thousand nodes of my network each of them might have to send a boat to each of the others that's a million votes going over the network and then each of them might have to send a receipt to each of the others for each of the votes it received that's a billion votes going over the internet and then that's one round of voting then each of them has to look at their votes to make some decisions and then send out new votes and that's a new round and so you have multiple rounds of these billion messages and then that's just to decide a single yes/no question but that doesn't tell you what order the transaction should be in if you have a bunch of transactions and so then you have to do a system where everybody proposes an order which is even more inefficient and not fair because whoever ends up winning that election put the transactions in whatever order they wanted maybe they were bribed to put them an unfair order that wasn't really realistic according to when the transactions were truly performed and it doesn't have fair timestamps and you might have been bribed to leave some transactions out of your proposed list so you don't have fairness and you have horrible inefficiency and you have lots of problems with the pure voting based systems however they are the gold standard in not the blockchain community in other communities they go back decades distributed systems fault-tolerant systems and you have these beautiful theorems of the FLP theorem the cap theorem all coming from this community and so what I would say is that a pure voting based system is probably not the best way to build a DLT but they are incredibly important for helping us understand the issues for helping us to understand the math and then allowing us to build systems like the virtual voting systems that are fast and have good math proofs and so they are maybe academic but they're useful academically and that is what a pure voting based DLT is

You May Also Like