There is no reason to be disappointed. The guarantees are same as before. That is, Its 51% computing power that sets the next reality not a central/political/technical institution.
51% being honest is fundamental assumption of a cryptocurrency [1]. The assumption gets stronger as total computation power increases.
[1] Its necessery but not sufficient condition to make the whole state dishonest. You would need to have >50% computation work done.
This assumes that the increases in computing power were not added via a mining pool, in which case the additional computing power is making things worse.
This is one reason that many view mining pools as one of the worst failures of bitcoin / other blockchains.
It's not total centralization, but it's a far cry from the level of decentralization that would be possible if mining pools did not exist.
There is a lot about this topic in the new, free book from Princeton authors titled "Bitcoin and Cryptocurrency Technologies" (a great read)[1]. There is much research being done on the front of finding the best way to eliminate mining pools or at least stop them from potentially centralizing consensus to just a handful of people.
[1]: https://freedom-to-tinker.com/blog/randomwalker/the-princeto...
Bitcoin might have 51% dishonest problem but it can be made honest again by increasing total computation. Whether 51% are honest or not is function of total computation power [TCP]. Probablity of 51% being honest goes to 1 as TCP goes to infinity.
My point was, only very few people are about something like this.
This is the whole "longest chain" versus "longest valid chain" distinction. A Sybil attack can censor the Bitcoin blockchain, but 51% doesn't gain full control of the network.