The stakers have to perform consensus network and block building actions continuously online, or they are penalised by losing value from their stake over time rather than gaining it. The penalty is reasonable so a small downtime during updates and such isn't a problem.
This incentivizes them to have reliable networking and hardware, running 24x7.
However they are heavily penalised if they are seen to do things like double-voting by accident, so they can't just put up duplicate systems and forget about it. High availability failover is something they can only do carefully.
This is different from the current proof-of-work miners. If a miner stops or screws up, they won't gain mining fees during the stop, but they don't lose anything either.
I understand all that but what I don’t understand is what encourages more physical nodes with high quality bandwidth because it seems like if me and you both want to stake for PoS then we can even be better off sharing the same hardware and buying higher quality hardware than we could individually, and this is a sub-linear incentive which I would think would result in the physical hardware footprint significantly shrinking. Am I missing something?
Now you might say well isn’t that a good thing for less resources but without an incentive to decentralize physically won’t there be a risk of centralization and potentially catastrophic outcomes for the network?
One factor fighting against this is that the validator inactivity fees spike to be larger if a lot of validators are offline at once. So if you think all the existing validators are in e.g. AWS us-west, you are taking more risk by putting your validator in AWS us-west too, because your failures are more likely to be correlated with other validator failures.
> what I don’t understand is what encourages more physical nodes with high quality bandwidth because it seems like if me and you both want to stake for PoS then we can even be better off sharing the same hardware and buying higher quality hardware than we could individually.
For the same reason that there are a number of mining companies too. But you're right, PoS doesn't 'fix' that problem.
This incentivizes them to have reliable networking and hardware, running 24x7.
However they are heavily penalised if they are seen to do things like double-voting by accident, so they can't just put up duplicate systems and forget about it. High availability failover is something they can only do carefully.
This is different from the current proof-of-work miners. If a miner stops or screws up, they won't gain mining fees during the stop, but they don't lose anything either.