> Rewards are given for actions that help the network reach consensus. You'll get rewards for batching transactions into a new block or checking the work of other validators because that's what keeps the chain running securely.
The unstated gotcha here is that the chain operates through a variation of BFT agreement where staked coins vote for new blocks. All the usual BFT constraints apply -- namely, if fewer than 66% of the staked coins can reach a quorum, the chain stalls. This would mean that the network is only as resilient as the nodes that contribute the least-resilient 33% of the coin votes.
I bring this up because it has some pretty terrible resiliency implications below.
> Although you can earn rewards for doing work that benefits the network, you can lose ETH for malicious actions, going offline, and failing to validate.
If the coin is successful, then this really serves to incentivize DDoS attacks.
PoS is fundamentally a "rich-get-richer" system -- the means of making new coins tomorrow are intrinsically tied to owning coins today by the protocol itself. There's no way around this.
This intrinsic coupling has two significant economic implications, which in turn impact the chain's resiliency:
* The price of coins is a function of their expected future revenue from staking. If you want to become a staker, the expected value you'll pay for your coins will include not only the spot price of the token, but also all the future (but time-discounted) revenue that coin will earn you from staking it. This is the lower-bound case, too -- if there's something even more profitable than staking you could be using the coins for, then the expected price of the coin will reflect that activity instead of staking.
* Coins minted from staking would need to be continuously re-staked in order to maintain current profitability relative to your competitors. Stakers can't afford not to do this, unless there's a buyer (or use-case) that can give an even higher ROI than all future ROI from staking block rewards.
This impacts chain resiliency as follows:
* There will never be any spare capacity for fail-over, since staking your coins will always be more profitable than keeping them on stand-by in order to recover from failed block producers. If the chain encounters a liveness failure (e.g. more than 33% of the staked coins go offline), new block producers can't just step in and take over without buying the coin first (and the more successful / long-term-valuable the coin is, the higher that price will be). The price of the token would either need to first come down to whatever level the highest-bidding poor but honest block producer can afford, or the protocol would need to slash a portion of the 33% of offline coins until quorum can be met. Neither of these things is instantaneous, so you'd be looking at a dead chain for some non-negligible amount of time. (Note that this is not true for PoW -- mining rigs can lie dormant until the difficulty falls to the point where it's profitable to turn them on, thereby ensuring that a liveness failure in the profitable block-producers does not lead to an overall liveness failure for the chain).
* It's very costly for new honest block-producers to join the network, which will make the network brittle. As the token becomes more successful and its long-term value realized, honest block producers will have to pay more and more up-front capital costs to start participating. The only people who can sell them coins to stake are their competition, so there's little seller information asymmetry to exploit here -- the seller knows exactly how much these coins are worth to the buyer for staking, so they will always price that in. This is a direct consequence of the means of coin production being tied to coin ownership.
* If the coin is successful, then there will reach a point where the cheapest way for a block producer to grow their revenue from staking is to knock other staking nodes offline (or hack them and cause their coins to be slashed from bad behavior). There's no fail-over capacity to take over block production if the staking coin quantity decreases, so the attacker only needs to succeed once in getting their competition slashed. Then, their fraction of all coins staked will increase due to other coins getting slashed. (Contrast this to PoW, where the attacker not only needs to knock the competing block-producer offline, but keep it offline indefinitely).
It would be a mistake to say that Ethereum 2.0 is in any way similar to Ethereum today. The economics of its PoS system make it more brittle, less resilient, and less egalitarian than Ethereum 1.0. Moreover, it would be misleading to say that Ethereum 2.0 is an open-membership system -- the economics make it so newcomers can't compete with the initial block-producers unless the price of the token always goes up faster than the initial block-producers believe it will.