Practical endgame on issuance policy

[aelowsson]

@OisinKyne

it does not fundamentally matter who runs the validator

Well then imo this might be where a model can be harmful rather than useful . It matters, at least to me, who runs the validator. If you just assume the staking providers raise their prices, you end up like this situation , with only the centralised entities remaining to offer services.

A misrepresentation or misunderstanding. The full statement is:

@aelowsson

For the purpose of calculating the effect on welfare, it does not fundamentally matter who runs the validator. The staking service provider will pass on its costs to the delegator in the form of a staking fee. The delegator will make the decision to delegate stake based on the full cost (staking fee and, e.g., assigned risks), and the principle applies just the same.

The topic under discussion is the welfare gain from a cost reduction for the ETH token holder. It is desirable to be able to calculate this cost reduction, and it turns out that we can simply specify the cost as the reservation yield of the stake. The SSP will pass on its costs to the delegator, who will decide to stake when accounting for the fee of the SSP. We can thus rely on the integral of the supply curve for computing the aggregate cost. This is a separate topic from how the composition of the staking set is affected by a change in issuance policy. In this case, it is certainly important who runs the validator, as addressed throughout my responses and the FAQ. Indeed, this is the reason for why a stake capping strategy should potentially be avoided in my view.

@OisinKyne

In this FAQ, it doesn’t make any reference to ‘home stakers’ (for the purpose of discussion, assume one or a group of people putting up < 32 eth for a validator, topped up with delegated stake, not a future orbit thing, or a 32 eth DV cluster of their own capital. Think rocketpool minipool squad). Where do they fit in with respect to your view of Ethereum’s goal below: (which i find far, far, too low of a goal for that matter)

I use the term solo staker to cast a wide net. For the sake of analysis, it is not critical if we are dealing with a squad staker or a small solo staker under Orbit SSF. My perspective was outlined in my previous answer.

@OisinKyne

The full cost of operating stake is best approximated as the reservation yield associated with the stake, that is to say the lowest yield at which that ETH would be supplied as stake.

Idk, I think the best way to approximate the cost of operating stake is primary research coupled with a model informed on this and other data (see later in post). The lowest yield at which Eth would be supplied as stake is a hypothetical/counterfactual we can’t answer as far as I know.

This is a misunderstanding of the topic being discussed in this part of the response. The topic is that the supply curve represents the cost of staking. For a better understanding, read my response to Keyneom starting here:

@aelowsson

The supply curve indicates the amount of stake deposited D at various staking yields y. It captures the implied marginal cost of staking. What that means is that every ETH holder is positioned along the supply curve according to how high cost they assign to staking, with their required yield implying that cost. Relevant costs include hardware and other resources, upkeep, the acquisition of technical knowledge, illiquidity, trust in third parties and other factors increasing the risk premium, various opportunity costs, taxes, etc. The area above the supply curve indicates the stakers’ surplus (what they actually gain) and the area below the supply curve the costs assigned to staking (the marginal staker would not stake at a yield below the supply curve).

Also review the associated figure.

@OisinKyne

The supply curve thus implies the marginal cost of staking, and the aggregate cost of staking becomes the integral of the supply curve.

You are arguing that the supply curve should dictate the cost of staking, I’m saying that would likely be bad for the quality of security of the network.

Also a misunderstanding of properties of the supply curve. Refer to my previous answer for an explanation of the topic being discussed.

@OisinKyne

I don’t understand

1. why you don’t think yields can go ‘close to 0’ in your proposals (at least thats what i infer from the framing of the concern only ‘being valid’ in this case. and
2. “No SSP can reasonably outcompete all others at an equilibrium staking yield of, e.g., 2% at 30M ETH staked” why do you say this? And why does this impossibility of out-competition not hold at 34m and 2.5% like the current day?

1. The supply curve will go almost vertical at a high proportion staked. Think of all the people that wish to hold ETH on hand, that will find a 0.5% staking yield too low if they need to trust a third party with their ETH, etc. Or review the cryptocurrencies that do not reach the full supply staked even at a 5-10% yield.
2. It holds at both 2% and 2.5%, that’s the point. A 0.5% issuance rate becomes a 2% issuance yield at 30M ETH staked. The discussion in that part of the text relates to the common suggestion during the debate at that time that striving for 30M staked is too low because then Coinbase may rely on economies of scale to outcompete all other SSPs. But this could only happen under a stake cap that targets a low stake. It is not a possible scenario under a more moderate reward curve because the staking yield would still be 2%.

@OisinKyne

to me this sentence reads like we should be on the same page, that perfect competition is not a reasonable assumption, so we shouldn’t adopt a curve that tends towards maximal marginal competition and thus imperfect markets like oligopolies and monopolies.

Perfect competition is a term in economics, and my suggestion in this part is that staking services are not perfect substitutes:

@aelowsson

Each SSP, reaching for a specific market segment, incurs unique costs besides the cost of running staking nodes, with a wide variety ranging from compliance to software. Indeed, CEXes have somewhat of a local monopoly on their customers-as-delegators, and the opportunity cost of keeping fees competitive with any onchain option is presumably so high that it does not represent the profit-maximizing strategy. What is clear is that competition for delegated stake will unfold across diverse market segments, and perfect competition is not a reasonable assumption.

One take-away is that fees will not be driven to zero.

@OisinKyne

That is fair and I didn’t realise you were an author on the (V)Orbit work, I apologise.

Thanks.

@OisinKyne

Why the delineation between 1 and 32 and not a different level?..

I focus on 32 ETH as a natural baseline for historical reasons. It would be particularly painful to make many 32-ETH validators unprofitable from a decentralization perspective. I focus on 1 ETH as a natural lower bound in Orbit SSF designs.

@OisinKyne

They do not make Ethereum cheaper, but inflationary, or ideally, dis-inflationary regimes would be better for the use of Ethereum than structural deflation.

Counterarguments here.

@OisinKyne

…there is a long way to go in my eyes, before the Ethereum operator set becomes sufficiently decentralised for what it needs to be to be black swan secure, and I think bounding issuance, and making it even harder for marginal stakers to stay afloat, making distributed validators particularly economically non-viable at retail scale, will push Ethereum down the road I outlined in “the staking problem”, which was more eloquently framed by Ameen, as USDC fork choice rule.

Great link! Interesting to read thinking from the early days. A relevant section to me would be this.

They’re violating the rules they are supposed to abide by, but with 92% of all ether, who’s going to stop them? “This is important”, they say. Within minutes, a new longer chain has outpaced the fork that contains the hack, history has been rewritten, the world has collectively decided to forget those transactions.

While it is not perfectly clear if the author meant to refer to all ETH or all staked ETH, a key reason for MVI is precisely because of the fear that all ETH would be staked when some black swan event takes place. Consider the impact of a black swan event causing a chain split in this scenario. It would make it very hard for the community to come to social consensus. I’ll quote my write-up from the MVI thread that outlined this idea:

40. …LST holders and any application or user who needs the LST to preserve its value will develop a shared destiny with both the LST and ultimately the LST issuing organization (the SSP).

41. In the case of a mistake or misdeed, which may also take place at any mechanisms designed to regulate the organization (such as smart contracts, on-chain governance, or government regulations), Ethereum’s social layer may not have the capacity for appropriate (non) measures.

42. It would require Ethereum to destroy a large part of itself. The affected users may prefer to reinterpret the mistake or misdeed as something entirely different. Once you become the money of Ethereum, you to some extent become the social layer.

43. The money function is in this way a Ryanian “stratum for cartelization” that acts one layer above the various strata (e.g., MEV extraction, block-timing manipulation) that Ryan explores in his post on the risks of LSTs.

44. We are no longer only concerning ourselves with the proportion of the staked ETH under an LST, but the proportion of the total ETH under an LST. The corrupted institution(s) correspondingly also sits one layer above the consensus mechanism, namely the social layer.

45. It became apparent with The DAO that if the proportion of the total circulating supply affected by an outcome grows sufficiently large, then the “social layer” may waver on its commitment to the underlying intended consensus process.

46. If the community can no longer effectively intervene in the event of for example a 51 % liveness attack, then risk mitigation in the form of the warning system discussed by Buterin may not be effective.

47. The consensus mechanism has in this case through derivatives grown so large and interconnected that it has overloaded its ultimate arbitrator, the social consensus mechanism. It is a special and sort of inverted case of issues Buterin warned about.

If the deposit ratio grows very high, it is reasonable to argue that Ethereum becomes less secure. We can avert this by reducing issuance.

@OisinKyne

I am unconvinced that bounding issuance to d rather than D won’t cause ultra low cost validators driving the equilibrium APR to way lower than most seem to expect, resulting in a mono-culture of ~3/10 entities that can make profit at such low rates.

The switch to d from D will be done to preserve the same APR at the same proportion staked when Ethereum undergoes either deflation or inflation and follows analysis of Ethereum’s circulating supply equilibrium. It will not lower APR in the way you fear.

---

I will not have time to provide more answers going forward. I do find it desirable with an equilibrium staking yield above 1%, and also see this as the likely outcome from a change. But I also consider it a problem if the quantity of stake grows very large, which is why I think we should specify a lower issuance yield in this scenario, to stop the growth. This is a matter of trade-offs, where we must seek to balance many different priorities.