A while back I posted some doubts I had about the idea of 12 witnesses in byteball.
The byteball jesus video (which I thought to be a slightly unfair dig at Roger Ver) nether the less had some great comments, one of which I repost here. Quote from Surirelav
This is exactly the kind of discussion I was looking for..
12 seems like a low number but a witness in the Byteball system ideally is not just a single person but someone who represents a big business or institution with a lot of real world value. Let's take Jeff Bezos as an example. He has so much to lose in the real world that it would be very hard to persuade him to collude with 6 other very high profile people of high market cap companies/high profile institutions across the globe. Let alone, abduct and torture 7 of them before anyone finds out and can replace them.
All the assets of a witness in the real world, its entire reputation, all that is a security bond, not just what it has or what it earns in the internal economy. This makes the witnesses anchors to the real world.
From a game theoretic point of view I would argue that a 12 witness system like this is actually more secure than a PoW based one. If you up the number of witnesses it will be much harder for users to know and recognise them all so they would just go with the "default" and not care about it much. So you have to have a number that is both safe and practical, 12 seems to fit the job. This number is large enough to protect against occasional failures but is not too large so that users can still know every witness.
And we want the witnesses to be accountable, that’s why there is a way to replace them. The way it works is that every user, with every transaction, posts a list of 12 witnesses he trusts. But we won’t have consensus if the lists posted by different users are too far apart. That’s why the lists of neighboring transactions on the DAG must differ by no more than one position.
The fact that witnesses can be replaced is another difference with miners. The witnesses are accountable. If a witness loses trust of users, users immediately signal their mistrust by changing their witness lists.
What miners do, they look at your transactions and decide, “I like this transaction and will include it, or I don’t like this transaction and won’t include it”. With witnesses, it is the reverse. It is users who look for witnesses to determine the consensus order of transactions. Witnesses don’t decide anything about individual transactions. They are expected just to be there.
What we do know from 9 years of Bitcoin / PoW though is this:
Miners are gatekeepers; they have the power to include or not include your transaction. So the access to the ledger is not quite free, you have to meet the miners’ criteria for your transaction to be included in the next block.
Miners are not accountable; they are supposed to make money, period.
Fee market is an uncertainty; When the blocks are full, there is competition to get into the next block. Paying a higher fee supposedly increases your chances to get into the block. But a payment system which is supposed to be used in everyday business, is not a game of chance, there must be guarantees that once you send a transaction according to all the rules, which is properly signed, spends existing money, and is not a double-spend, after you send it, you can immediately disconnect and still be confident that it goes through. In blockchains, when the blocks are full, there is no certainty, no guarantees, no such thing as the right fee.
Trade-off between security and cost; block reward is supposed to be reduced over time and eventually go to 0. This means that miner incentives from block reward will decrease and they should be replaced by fees. Which means that the system is as secure as the total amount of fees paid, and there is an uncomfortable trade-off between security and cost.
Negative balance of resources; miners burn energy and have to pay to energy companies, outside the ecosystem. Which means that the balance between resources flowing in and out of the ecosystem is negative.
Outside incentives are ignored; lastly, miner incentives are only inside the ecosystem. They make more money by mining honestly than by doublespending Bitcoins, that’s true under assumption that all their incentives are inside Bitcoin. If we forget about the outside world, as if it doesn’t exist. But in reality, outside incentives do exist. Even more, if there are other tokens issued on the same ledger, as it is the case on Ethereum, the block reward is still in the native currency only but now this should be balanced against doublespending not only the native currency but the other tokens too. If the tokens are worth more than the native currency, the system becomes insecure.
So it does make sense to try out other consensus mechanisms. I really like Bitcoin a lot, but I love Byteball for trying something new.
