I guess what I'm asking is if things like the hard fork of The DAO Ethereum blockchain are actually a fundamentally solved problem. Stuff like that happening when our banking infrastructure has moved to a blockchain sounds like a disaster.
I guess what I'm asking is if things like the hard fork of The DAO Ethereum blockchain are actually a fundamentally solved problem. Stuff like that happening when our banking infrastructure has moved to a blockchain sounds like a disaster.
Having >50% hashing capacity allows you to arbitrarily select which transactions will or won't be confirmed.
Additionally you can perform double-spends, that is the most important thing (in my opinion) that proof of work was supposed to prevent. Of course with more confirmations this attack becomes more and more expensive to execute. For details see Majority Attack [0] page on Bitcoin wiki.
On the other hand, the DAO fork was a result of centralized control due to social factors. As with bitcoin, a >50% consensus is necessary to fork a blockchain. But because there is a single overwhelmingly popular ethereum client, the developers had the ability to make the software obey the fork by default. A referendum was conducted in good faith, and the devs flipped the switch. But again, this was only possible because everyone was using the same client to begin with, and few people bother to change the default behavior.
I myself doubt that segwit will ever get adopted in bitcoin. It looks like even if a small group of actors want to block some change, they can. So that will mean that the protocol will remain pretty much the same with the good and the bad features.
I'm not very well-versed in computational power requirements on this scale, but it doesn't sound at all infeasible that a nation state (or a couple in unison) could setup a "51% attack", thus reducing the social form of control back to only the computational one. And if anything, that threat is only going to become worse with ever-increasing computational capability, advent of quantum computers at some point, etc. Nations (and other huge players) have the resources to tap into all that new power, while the individuals ideally making up the bulk of a blockchain's computing power would most probably not, at least not in big enough scale.
I don't see how that could be prevented, and that leaves me somewhat confused on why the blockchain is advertised as "immutable" as it is.
The ethereum ecosystem has placed a lot of trust in the ethereum foundation, plus the foundation controls a large percentage of the fully validating nodes, and is in close contact with all major miners, exchanges, etc. For this reason they are much more able to push through contentious changes. This is provably not the case in Bitcoin, as demonstrated by the active segwit vs. hardfork political deadlock.