pull down to refresh
171 sats \ 13 replies \ @047b99aa4d 10h \ on: A soft fork is a 51% attack on Bitcoin bitcoin
So expand on that. What happens to the nodes running the new software IF less than 51% of the hashrate isn't mining blocks using the new software? Mechanic has described some kind of mechanism where the chain split can eventually catch up to the other and then a wipeout occurs. Any thoughts on that scenario or how/if it could happen?
Let's imagine a fork happens and 40% of hash is on the new rules and 60% of hash remains on the old rules.
The chain with the old rules outgrows the one with new rules. (for now)
Let's say at a later point in time more miners switch from the old rules to the new rules. Now 60% of hash is on the new rules, while 40% is on the old rules. In this case the new chain will eventually outgrow the old one. Now the new chain is also the valid one according to the old rules, since it is longer. Thus the blocks on the old chain are replaced by the ones on the new fork.
reply
That’s a correct description of what would happen if the hashrate moved that way, but an unlikely scenario.
It’s the best practice to only start enforcing soft fork rules if they are expected to be enforced by a majority of the hashrate, exactly because a minority-enforced soft fork leads to a chain split. It would be against the interest of the miners on the heavier chain to switch to the minority-enforced soft fork chain, as they would be sacrificing their block rewards. Similarly, starting to enforce a soft fork with a minority of the hashrate most likely means that the block rewards will never become part of the best chain and therefore likely to incur the whole cost of mining with no reward.
In effect, anyone proposing to activate a soft fork with a minority chain is essentially advocating to create a forkcoin.
reply
So this is the wipeout scenario. If only Ocean is mining blocks on the new software split I don't know how they think their chain would ever be able to catch up... Also, how do block subsidies work in the "split then catch-up" scenario? If there is a reorg after 100 blocks pass does it invalidate the 3.125 Bitcoin awarded?
reply
Anything in the blocks that get reorg'd out is gone, so that includes block subsidies.
reply
Anything in the blocks that get reorg'd out is gone, so that includes block subsidies.
I think your "anything" is overly general; transactions that are valid on either chain will probably get "sniped" across. The miners restoring these on the new chain would get paid fees from the transactions, and the people who had originally broadcast them would probably even be relieved that their transactions had returned to confirmed status.
reply
Good point! I hadn't considered replay protections and cross-split sniping. That makes it a good deal more complicated than I considered.
reply
It’s a frequent question we get on Bitcoin Stack Exchange: if a block gets reorged, what happens to the transactions in that block? Do they all get unconfirmed and returned to the mempool?
And the important thing to realize there is, that from the perspective of each chaintip the other block might as well not exist. The best chain has only one block at each height, so competing blocks do not at all influence each other in regard to what is included.
Most of the time two competing blocks at the same height will include almost exactly the same transactions with some minor differences in what txs they had seen when they created their template, or some locally prioritized transactions. Obviously, the coinbase transactions will differ, and very occasionally, one block could include a replacement of a transaction while the other block included the corresponding original. But because miners try to maximize the revenue from the available mempool, they pick all the same transactions, and for the most part, either both blocks or neither confirm a transaction, so it makes no difference for the users which block wins out in the end.
Addendum:
Obvious exceptions are of course when someone is deliberately reorganizing the chain to revert a previously confirmed transaction, or when a reorg is so deep that coinbase transactions matured on both chaintips and said coinbase outputs don’t exist on the competing chains, so a reorg could remove whole chains of transactions from the history.
reply
the new chain will eventually outgrow the old one
This makes sense. Thanks for explaining it. The important point seems to be that a fork needs sustained period of majority hash power enforce its rules.
Thus the blocks on the old chain are replaced by the ones on the new fork.
This comes with a reorg of some length because the old chain that miners were working on gets orphaned.
reply
Was the double negative on purpose there, or did you mean “if less than 51% are enforcing the new rules”?
There are just two cases:
- If a majority of the hashrate enforces the soft fork, the non-enforcing minority may occasionally find a block considered invalid by the majority, but the soft fork enforcing majority will ignore such a block and eventually build a heavier competing chaintip at which point the minority will reorganize to the majority chain.
- If only a minority of the hashrate enforces the soft fork, it will create a permanent fork as the minority will not accept the non-enforcing majority chaintip and the majority chaintip will accumulate more work faster. The majority chaintip will simply pull away and the enforcing miners will be stuck on a less-work chaintip.
reply
Yeah, that was a mistake with the double negative. Your response clarifies what I was after though. Appreciate it!
reply
To be honest, I find this stuff difficult to reason about, so here's my best guess:
If a soft fork doesn't have majority hash power enforcing it
and if a block gets mined containing a transaction that is invalid according to new rules
miners running new rules won't build on that block
If majority of hash power will build on that block
chain will split
and miners and nodes enforcing new rules will fork off from rest of chain
I don't see how the soft fork side of the split is ever reconnected with old rules side.
reply
That’s right, the only way the network would converge back on a single chaintip without intervention would be if the soft forking chaintip outpaced the non-enforcing chaintip and the non-enforcing nodes reorged to the soft fork chaintip.
This is highly unlikely when only a minority of the hashrate enforces the soft fork.
reply