Some incomplete thoughts on economics of 1) liquid staking; 2) restaking

No free lunch: yield dilution from liquid staking

Proponents of liquid staking point to capital efficiency: use your capital twice, once in staking, once in DeFi. Strictly better?

However as total amount staked increases, the yield offered is diluted. If liquid staking enables much more Ether to be staked, thus diluting yields, are prospective stakers really better off? Do I really prefer 2.5% (which gives me an LST I can deposit into DeFi) over 5% (if LSTs didn't exist)?

Restaking, or, keeping up with the Joneses

Most of the arguments I've heard about restaking draw incorrect parallels to leverage in traditional finance and warn of some "bank run" if the system delevers. That is an incorrect concern; when a user gets slashed on one AVS, his stake on all AVSes is reduced.

What I think actually a risk is a corollary of the yield dilution described above. If 1 ETH can be staked many times, it will be, diluting yields and forcing stakers to follow suit just to get a reasonable yield.

Simplistically, if (in the absence of restaking) 5 staking opportunities each have $200M earning 5% yield, but then restaking allows all $1B to stake all five opportunities, then in order to still earn 5% yield one would have to restake in all 5 opportunities. More risk for the same yield.

Does staking offer 'carrot' or 'stick' incentives?

Carrots are rewards for good behavior, while sticks are punishments for bad ones.

A naive view of staking would be to say that it offers carrot incentives: extra tokens for participating, no harm if you don't.

A more nuanced view is that staking actually offers stick incentives: as staking rewards are funded through inflation, anyone who does not stake is seeing their value inflated away.

For Ethereum in particular, the more that is staked across the network, the bigger the stick becomes, because inflation scales with the square root of total stake. From this perspective, liquid staking (which has the effect of increasing total stake) is actually making the penalty for not staking larger.

Side effect: risk of governance subversion

People react to incentives. The rational thing to do given the current incentives is to stake (or restake). And given the current protocol constraints (min 32 ETH to solo stake; no in-protocol delegation), the rational decision for many is to utilize the popular third-party solutions that offer this service.

The obvious problem is that this puts power in the hands of those third parties.

It reminds me a bit of an anecdote from early HFT days. Spread Networks built an optimized cable between Chicago and NY and sold the service to all of the major HFT firms. Prices were exorbitant but everyone paid, and no one was better off than before.

Liquid staking/restaking feel similar to me. By making it easier to earn yield through them, they increase the amount staked while representing an increasing proportion of the total stake. Protocol design needs to not actively incentivize usage of third parties.

The benefits of larger total stake

The good news, it indeed becomes more expensive for an outsider to accumulate enough stake to mount a 33%, 51%, or 67% attack. Though, if the attacker is not actually an outsider, but rather a major validator coalition, then this argument doesn't work as well.

Wat do?

• MVI to reduce the implicit cost of not staking, thus decreasing the total amount staked.
• Perhaps: in-protocol delegation to make the in-protocol option more competitive with the third-party option. However, the effect on solo staking rates needs to be studied first, since presumably some solo stakers would just delegate if it were an option.