Loading ..

Issuance Issues — Initial Issue

^ Playing with the polysemic nature of the word, issue, to mean both “a vital or unsettled matter” and “the thing or the whole quantity of things given out at one time.” sorry lol… couldn’t help it :)

^ do you think that is what Ethereum looks like? lil’ guy is out here doing his best. ◇🍴
⋅
by mike based on discussions with caspar & ansgar – saturday, march 30, 2024.

Disclaimer: This article only contains the author’s opinions and does not represent the view of any organization.

⋅
tl;dr;
Issuance defines how Ethereum pays for security. Entering mid-2024, the realities of staking have changed seismically since the inception of the beacon chain at the end of 2020. The future remains uncertain, but we can explore the impact of issuance on solo staking, ETH the asset, and protocols building in & around Ethereum consensus.

Unfortunately, we do not have the final answer for the staking endgame. As research in this direction continues, we must also be pragmatic about where ETH staking is today and how it might evolve over the next 1-2 years as we work towards an ossified protocol. With the upcoming Electra fork, we argue for making a conscious decision. Whether or not that involves changing the reward curve, intentionality is essential. Presenting the arguments for both sides is vital to help inform that decision.

We hope to be more expository than prescriptive. That said, we acknowledge our biases; we believe changing the issuance curve in the Electra fork is important and should be seriously considered. This article aims to make that case based on the available information, but we hope others express opinions on this topic – we view it as one of the most important conversations!
⋅
Contents
(1) What is issuance?
(2) What is the current situation?
   (2.1) How may the situation evolve?
   (2.2) … so what? What’s wrong lots of stake?
(3) What are our options?
   (3.1) Apples-to-apples: comparing staking equilibria
   (3.2) Consider the solo staker
(4) Summary
⋅
Related work

Article Description
Minimum Viable Issuance Anders’ first post
Properties of issuance level (part 1) Anders’ second post
Endgame Staking Economics: A Case for Targeting Caspar & Ansgar’s post
Electra: Issuance Curve Adjustment Proposal Caspar & Ansgar’s Electra proposal
UCC2: Ethereum’s Staking Endgame Jon, Hasu, Caspar, & Ansgar’s discussion
Initial Analysis of Stake Distribution Julian’s post
Reward curve with tempered issuance: EIP research post Ander’s third post

(1) What is issuance?

Issuance describes the rewards distributed by the Ethereum protocol to incentivize secure block production and validation. These rewards are minted as new tokens and given to network participants who conduct consensus duties. The total supply of ETH depends on the issuance and the burn, introduced in EIP-1559. The following timeline shows a brief history of Ethereum issuance; for more, see “The history of Ethereum”.

Issuance timeline

Since the beacon chain genesis, the Proof-of-Stake issuance has not changed. The curve (a parameterized inverse square root) calculates the amount of yield as a function of the amount of ETH staked.[1]
yield=2.6⋅64√staked ETH
The parameters in the numerator come from estimates signaled in this PR. The figure below shows this curve graphically.

^ source.

The parameter selection of the inverse square root was based on a rough estimate: aiming for a modest but reasonable 3.3% yield with 30 million ETH staked (see “Serenity Design Rationale” for more details on the choice of the inverse square root specifically). The primary feature of this curve is highly incentivizing at least 10 million staked ETH. Beyond that, no specific stake amount is targeted, with the rewards gradually decaying. Today, there are more than 31 million ETH staked; the 30 million ETH target may have underestimated the staking supply. Let’s dig into this a bit more.

(2) What is the current situation?

As of March 2024, staked ETH accounts for approximately 31 million of the total supply of 120 million tokens – a “stake rate” of 26% (see https://www.validatorqueue.com for the latest). The amount of staked ETH has only ever increased, even (and especially) after enabling withdrawals. The figure below shows the growth in active validators and ETH staked over the past three years.

^ Hildobby dashboard. Annotations from Endgame post.

We outline three potential factors contributing to the continued demand for staked ETH.
Staked ETH factors today:

  1. Risk – Perceived & real risks involved with staking have declined due to the continued stability of the network and the successful launch of withdrawals.
  2. Convenience – Liquid staking tokens, centralized exchanges, custodians, and wallet providers continue lowering the entry to staking.
  3. Yield – MEV rewards and airdrop farming from staking/restaking services provide additional yield for staked ETH beyond the rewards from the consensus layer. This yield is sometimes called exogenous as it comes from outside the protocol.

We are not in equilibrium, where the amount of staked ETH remains relatively stable; the number of new validators far outpaces the number of validators exiting the protocol. Given this reality, we must consider possible outcomes over the next few years.

(2.1) How may the situation evolve?

Predicting the future is hard; we can’t say if/when the new staking inflows will stop under the current issuance regime. The Cancun/Deneb fork included EIP-7514, which limits the maximum amount of validators that can activate or exit during each epoch at eight (MAX_PER_EPOCH_CHURN_LIMIT = 8). As a result, the expanding entry queue causes a delay in activation (now at four days) for in-bound validators. So long as this queue is non-empty, the maximum number of validators enter the protocol during each epoch. We postulate that the amount of ETH staked will rapidly expand during the next 1-2 years due to the following factors.

Factors contributing to the demand for staked ETH in the next 1-2 years:

  1. ETH price appreciating – Without over-indexing on the market trends, current prices indicate we may be entering a bull market. If ETH price goes up, demand for ETH-denominated yield increases because this rapid price appreciation results in higher USD-denominated yield.
  2. Restaking demand – With EigenLayer quickly attracting over $11 billion TVL, we expect demand for restaked ETH to increase. When AVSs launch, airdrop farming, and new yield opportunities will drive staked ETH yields up (in the short to medium term, just the potential for AVS token airdrops will induce demand, regardless of the actual yield produced by the AVS in the long run).
  3. Interest rates – At 5.33%, the current Fed Funds Rate is at the highest level since March 2001 (it was also around 5% in mid-2007 before the Financial Crisis). When this rate declines, more institutional capital will be looking for yield, and it seems reasonable to assume some of it will end up in staked ETH.
  4. LST/LRT demand – Liquid staking tokens and liquid restaking tokens are a blessing and a curse. While they lower the barrier of entry to staking and aim to help decentralize the validator set, they also induce massive demand for staked ETH. Like AVS tokens, farming LST/LRT airdrops will likely cause large inflows into these tokens in the coming years.
  5. Staked ETH ETF potential – We have no idea if an ETH ETF will launch, but we know the demand for staked ETH ETFs is high. Fidelity recently amended their ETH ETF filing to included staking, and it seems like the ETF issuers will push hard to offer this product. While this is great for the legitimization of ETH, it results in potentially massive staked ETH inflows by presenting the option to institutional capital allocators.

To summarize, the current amount of staked ETH already exceeds the rough heuristic used to construct the issuance curve in the first place, and there are many reasons to believe that the demand for staked ETH will increase in the next 12-24 months.

(2.2) … so what? What’s wrong with lots of stake?

The potential for consistent staking demand begs many questions about the resulting high-stake-rate regime. Many Proof-of-Stake chains already have much higher stake rates than Ethereum.

There is nothing fundamentally broken about having a majority or super-majority of the native token supply staked (it is worth noting that many Proof-of-Stake chains have a much more concentrated distribution of the tokens; early investors and employees may prefer staking large portions of the supply because that supply is illiquid (due to lockups) but the yield from the staking is liquid and salable, cashing out some of their investments). Additionally, some chains (Solana being the most relevant example) don’t have slashing rules defined in the protocol; risk-free staking is an attractive proposition.

At first blush, massive amounts of stake seem like a good thing. More stake implies more economic security for Ethereum; “the more, the merrier,” right? While we think Ethereum (the blockchain) would operate fine (from a technical aspect) in a high stake-rate regime, we also believe in serious side effects to ETH (the asset) and Ethereum (the community and governance layer) resulting from a majority or a super-majority of ETH being staked.

Again, we reiterate that none of the above is guaranteed. As with many parts of the Ethereum protocol, we are trying to perform a worst-case analysis of the situation to make Ethereum more anti-fragile. We also bring a healthy understanding of the rate at which Ethereum updates have occurred historically and the amount of time it would take to respond directly to any of the above side effects. Lastly, these problems mostly do not manifest in a single event but belong to the more pernicious “death by a thousand cuts” variety.

(3) What are our options?

The recent UCC2 episode with Caspar and Ansgar frames it nicely. As we look to the Electra hard fork, we must decide if we want to change issuance or not. Changing the issuance curve should only be done with careful consideration. But on the other side of the same coin (or token?!), not changing issuance should only be done after careful consideration. Indecision is a decision; inertia is real. Defaulting to doing nothing may be embracing a difficult-to-reverse trend.

Counterintuitively, we believe that not changing issuance in Electra is a more committal decision than changing it. The following thought experiment tries to illustrate this idea more plainly. Consider the following three outcomes for the stake rate:

Without making a value judgment on which outcome is best, we claim that moving in the direction of increasing stake (Outcome #1 -> Outcome #2 -> Outcome #3) is much easier than moving in the direction of decreasing stake (Outcome #3 -> Outcome #2 -> Outcome #1). Defaults are sticky, and after paying the upfront cost of running a staking operation or deploying capital, individuals or institutions choosing to unstake must overcome this sunk cost. Consider potential timelines for the two subsequent hard forks:

We use these timelines as pure hypotheticals, but the back-of-the-napkin calculations show that if we do nothing in Electra, it is highly possible that we land in Outcome #2 by the time of the F-star fork. As discussed above, we expect moving from Outcome #2 -> Outcome #1 to be quite difficult. Thus, we likely constrain the set of endgame staking outcomes by not adjusting issuance in Electra. Again, this might not be how it plays out, but we should go into the decision clear-eyed. Delays in hard forks are possible, and longer time horizons raise uncertainty. Maximizing for optionality is a guideline that might make sense to follow. As time progresses, the community may agree that a 100% stake rate is the correct endgame outcome; that option would remain fully feasible even with an issuance change. Preserving optionality is the main reason to take action in Electra.

If we do want to begin down this path in Electra, Caspar and Ansgar have proposed a new issuance curve based on Anders’ extended analyses. The plot below shows the juxtaposition of the current & hypothetical curves.

^ source.

We defer the details of “real” versus “nominal” yield to the next post. It’s easy to look at this proposed change and have an adverse gut reaction; “Yields going down is bad for solo stakers and the ‘good’ staking operators of today” — that was my initial reaction too! But after digging in, I think it misses a few subtle details.

(3.1) Apples-to-apples: comparing staking equilibria

The main issue with the current discourse is that comparing today’s yields with the hypothetical yields under a new issuance curve is a classic apples-and-oranges situation. As underscored by Caspar in the pod, the true comparison is between:

  1. the real yield in equilibrium under today’s issuance curve, versus
  2. the real yield in equilibrium under a different issuance curve.

We use “equilibrium” to mean “the amount of staked ETH is not meaningfully changing.” The emphasis highlights that yields can only be measured against each other “once the dust has settled,” and the yield isn’t systematically changing; that is how we get the apples-to-apples scenario. Today’s rates are tricky to analyze because we are not in equilibrium (see (2) What is the current situation?). But yes, changing the issuance curve as proposed would reduce nominal and real yields at current staking levels. However, the apples-to-apples comparison considers the yields given a stable staking rate under today’s versus the proposed issuance curve. We believe the steady-stake environment is better under a reduced issuance curve because it could dampen the demand for newly staked ETH, preserving a non-zero real yield for those already staking. Once more, we defer the mechanics of this calculation to the follow-up post (and acknowledge that all of these equilibria have uncertainty because the staking supply curve is unknown). But this is the fair way to compare different issuance curves.

Ethereum designs around the solo staker. With the gold-standard solo staking community (see EthStaker), preserving the viability of this core pillar of the ecosystem is an absolute must. With that in mind, let’s consider how the issuance discussion impacts the solo stakers.

(3.2) Consider the solo staker

There are high costs to becoming a solo staker today. We break these costs down along two different axes. First, monetary vs. temporal indicates whether the cost denomination is dollars or hours. Second, fixed vs. ongoing denotes whether the cost is a one-time investment or a continual fee to pay. The following is a rough sketch of the cost borne by a solo staker who runs a node.

Solo staking cost structure:

Notice that Hardware & Knowledge, the two fixed costs above, are high for solo stakers. The Stake & Maintenance costs are ongoing, but relatively lower (while acquiring the 32 ETH stake represents a high initial barrier to entry, the ongoing cost of the stake, i.e., the opportunity cost of staking rather than deploying the capital elsewhere, is relatively small). Based on these assumptions, we draw the following conclusion.

Conclusion #1: Given the current staking environment, most people who will solo stake already do.

While future upgrades aim to reduce the costs of solo staking, they are farther down the line. We believe the solo staking base is likely not growing significantly in the current regime.

Conversely, let’s consider the institutional staker that delegates their staking operation to Coinbase. Their cost structure looks pretty different. (Note: we ignore the smaller stakers who choose to delegate to Coinbase or Lido – this is a simplified example.)

Institutional staking cost structure:

Notice that Stake & Fees are both ongoing, constituting the entire cost structure of the institutional staker. This leads to our second conclusion.

Conclusion #2: The institutional staker is likely more price-sensitive to issuance than the solo staker because all the institutional staker costs are ongoing.

This conclusion is self-evident from the fact that most solo stakers’ costs are fixed and already paid. The institutional staker must constantly evaluate the value of staking compared to outside options because their costs are exclusively ongoing. We now arrive at the following (albeit reductionist) breakdown:

While this is an immense simplification, we are concerned with the directionality of the proportion of solo stakers, leading to our third conclusion.

Conclusion #3: An issuance reduction is not likely to reduce the proportion of solo stakers because solo stakers are less price sensitive than institutional stakers.

To reiterate, this is a simplification and intentionally ignores many factors; we don’t know how the staking endgame will play out. However, it is at least worth discussing an issuance change and its impact on solo stakers based on the rough sketch of the cost structures laid out above.

(4) Summary

Summarizing each section, we have:

  1. Issuance is how the protocol pays for security. The Proof-of-Work block rewards were modified twice, while the 2019 selection of the Proof-of-Stake issuance curve aimed for the rough target of 30 million staked ETH.
  2. There are currently 31 million ETH staked and a steady flow of inbound capital. Many factors that have arisen since the 2019 analysis selecting the inverse square root issuance may contribute to sustained inflows of ETH in the next 1-2 years.
  3. Changing the issuance in Electra may preserve the optionality of outcomes for endgame staking, while not changing it may result in a difficult-to-reverse high-stake-rate regime. Further, comparing issuance curves makes sense only in equilibrium. It seems like an issuance reduction preserves the viability and proportion of solo stakers.

That isn’t the end of the story, but the beginning. Happy to be on this journey with you all — “We’re all in this together.”

— made with ♥ by mike.

  1. Derivation: (mostly leaving this here so I don’t have to re-derive it; forgive the notational abuse). From the spec, we have the per-epoch reward in Gwei for a validator with a 32 ETH effective balance given S total stake as:
    Re=32×109⋅64√S
    With 384 seconds per epoch and 31,536,000 seconds per year, we have 82,125 epochs per year. Thus, the yearly reward in Gwei for a validator is:
    Ry=32×109⋅64⋅82125√S=1.68192×1017√S
    Thus the annualized yield (as a proportion) is:
    yield=Ry32×109=1.68192×1017/√S32×109=5.256×106√S
    Converting the units of S to ETH instead of Gwei, we have:
    yield(SE)=5.256×106√SE⋅109≈166.21√SE≈2.6⋅64√SE ↩︎

Last changed by  
 
michael.neuder@ethereum.org
3385

Read more

L1 R&D session: Proposer-Builder Separation

[Required] Summary: We will provide an update on the state of the research on Proposer-Builder Separation (PBS). The session will be divided between high-level design space discussions as well as specifics of EIP-7732. [Required] Facilitator: Terence and Mike. [Required] Note Taker: Terence and Mike. [Required] Pre-Reads: EIP-7732 Why Enshrine PBS

Nov 10, 2024
Issuance Issues — Subsequent Soliloquy^ real ones will know that Soliloquy is probably the best green run at Copper Mountain :)

^ p.s. how it feels to stare at these curves for too long... ^ p.p.s. did anyone else notice how much this ghost looks like Pearl from Finding Nemo?! ^ p.p.p.s. i just realized that Casper was probably named so b/c it was paired with GHOST?! 🤯🧐 $\cdot$ by mike – saturday, may 11, 2024. ^ happy bday Sabrina – hope you didn't take deja vu too personally (despite it being abjectly personal). $\cdot$ tl;dr; Hello & welcome to the second installment of the Issuance Issues.® As the previous iteration was so, very, very, very, very, well-received, I thought, "You know what people probably want more of? My thoughts on issuance!" Well, to quote the inimitable Bob the Tomato, "Have we got a show for you!"

May 12, 2024
Consider the ePBS

$\cdot$ by mike – february 7, 2024 $\cdot$ tl;dr; – Yesterday, after nine missed slots due to a bug in an optimistic relay, there was significant confusion around the role of relays and how ePBS would have changed the situation. This doc aims to reduce that uncertainty. The three important points to highlight: ePBS does not solve for builder bugs (the slots would've been missed regardless), best case ePBS – the proposer payment is faster and trustless, and worst case ePBS – bypassability reverts us to the same trust model as today. It follows that the only reason yesterday's event should change your mind about ePBS is if you think the proposer refund was too slow, and even then, you must acknowledge that in the worst case, the same refund process would have to occur in ePBS.

Feb 8, 2024
Inclusion lists: execution, consensus, & engine spec overview

$\cdot$ by mike based on discussions with Potuz, Terence, Francesco, Hsaio-Wei, Danny, Tim, Roman, Dan, Alexey, Lion, Mikhail, & Matt. $\cdot$ february 7, 2024 $\cdot$ tl;dr; -- This document presents a minimal view of what a potential spec for EIP-7547 could look like. We outline changes to the execution specs, consensus specs, and execution APIs to clarify how the pieces fit together. These changes are in draft mode; the goal of this document is to (i) convey the relative simplicity of the changes, (ii) highlight the key design considerations that came up in discussion.

Feb 7, 2024
Read more from michael.neuder@ethereum.org
Published on HackMD