I’ve been watching Ethereum’s scaling conversation for years, and every so often a proposal lands that feels different. This one does. A draft submitted just yesterday outlines a way for multiple transaction claims to share a single proof while keeping their payments and signatures completely separate. The idea is simple on the surface, yet the implications for gas costs and builder incentives run deeper than most people realize.
What This Ethereum Draft Actually Changes
At its core the proposal activates the LEANSTARK_SCHEME that was reserved earlier and adds clear rules for sponsored batches. A sponsor can pay for the retained batch leaf, which is the shared proof component holding several claims. Each member transaction still pays only for its own claims. That separation matters. It means a builder can drop some members from the group and still include the valid subset without forcing anyone to resign their transaction.
Think about the practical side for a moment. You submit a transaction that is ready for inclusion. Other people in the original batch fall out for whatever reason. Under the new rules you do not need a fresh signature just because the group changed. Your nonce and your receipt stay attached to your own transaction. The proof work is shared, yet the individual records remain intact. That design choice feels carefully considered.
How Sponsor And Member Payments Split
Payment matching happens during finalization. Recursive proofs authenticate the sources used in the batch. The draft draws a clean line between proving that claims are valid and confirming that the necessary payments travel with them. Discounted claims and sponsor payments go straight to the block builders. Ordinary single-statement proofs can still sit in the public mempool as usual.
I’ve found that this dual-track approach could reduce friction for users who simply want their transaction confirmed without becoming experts in proof aggregation. At the same time it gives sophisticated sponsors a clear economic role. Whether that balance holds once real numbers are measured remains an open question, but the architecture itself looks promising.
Building Directly On Earlier Work
The foundation is the earlier proposal that suggested aggregating cryptographic signatures and proofs outside the main execution process. Transactions would declare the statements they need verified. A recursive STARK would then establish validity at the block level. STARKs demonstrate that a computation ran correctly without forcing every node to repeat the work. Recursion lets one proof verify other proofs and combine their results.
That earlier work reserved the generic leanSTARK scheme rather than enabling it right away. The new draft switches it on and replaces a bare data-hash statement with a stable digest that represents a list of claims. The design continues a longer conversation about handling complex cryptography through aggregation rather than forcing every detail on-chain.
Sharing the expensive proof work while keeping individual payments and signatures separate could change how everyday users experience batching on Ethereum.
The Provisional Fee Numbers
The current revision lists some early gas figures. Native verification sits around 15,400 gas. Ordinary generic verification is listed at 572,000 gas. An intrinsic proof share is given as 5,100 gas, a leaf payment at 675,000 gas, and a batch claim at 4,200 gas. Leaf payments carry a separate charge because binding the payments adds fixed work beyond simple verification.
These numbers are still provisional. They need joint calibration, so they do not yet prove a final fee schedule or guarantee savings for users. Claims and payments themselves incur no extra intrinsic proof share. I keep reminding myself that early drafts often look expensive until the real measurements arrive. That is why the next section feels so important.
Resource Limits Still Under Discussion
Resource controls include a provisional block limit of 16,384 distinct dependencies. That limit sits independently of gas. A matching inclusion-list fit rule appears alongside it. A single leaf may contain up to 256 claims. A proof step may take 16 inputs. Neither figure is a hard block-wide ceiling.
These caps exist to keep the system predictable. Without them a single oversized batch could create headaches for validators and builders alike. The numbers will almost certainly move once real performance data arrives, yet having concrete starting points helps the conversation stay grounded.
Why Full-Block Measurements Matter
Before any activation the proposal requires measurements that cover the complete processing schedule. That includes late sponsor repair and per-proof charges that must fit inside normal transaction gas limits, overhead included. The benchmark material distinguishes measured individual-node and single-machine results from modeled multi-worker figures. Gaps still need to be documented.
An invalidated sponsorship could waste a builder’s effort. The draft is clear that inclusion gas would not reimburse that lost work. Concrete commitment construction and encodings stay inside the pinned proof-system specification. No new networking messages are introduced. That last point keeps the surface area of the change smaller than it might have been.
What Everyday Users Might Notice
If the numbers work out, users could see lower effective costs when their transactions ride inside a well-sponsored batch. They would still pay for their own claims, yet the heavy proof work would be covered by someone else. The experience might feel similar to how some layer-two solutions already abstract complexity, only this time the mechanism lives closer to the base layer.
Of course nothing is guaranteed. The draft remains under active development. No EIP number has been assigned yet, and no mainnet activation date exists. The submission supersedes an earlier pull request, but discussion threads and formal numbering are still ahead. I’ve learned to treat every early draft as a starting point rather than a finished product.
Potential Effects On Builders And Sponsors
Builders receive the discounted claims and the sponsor payments. That creates a direct incentive to include valid sponsored batches. At the same time an invalidated sponsorship leaves the builder with sunk costs. The risk-reward calculation will depend heavily on how reliable sponsors prove to be in practice.
Sponsors themselves gain a new role. They can cover the shared leaf while individual members handle their own claims. In my view this opens the door for specialized actors who focus on efficient proof aggregation and then sell that service to users who simply want cheaper inclusion. Whether that market develops smoothly is another question the community will have to answer through testing.
Keeping Individual Records Intact
One detail that stands out is the preservation of nonces and receipts. A nonce marks a transaction’s place in an account’s sequence. A receipt records the execution result. Sharing the proof does not merge those records. Each transaction keeps its own identity. That choice avoids a class of headaches that could have appeared if the batch had forced a single shared identity.
Perhaps the most interesting aspect is how cleanly the design separates the cryptographic heavy lifting from the economic accounting. Validity is established once for the group. Payments are checked per member and per sponsor. The two concerns meet only at finalization. That separation feels elegant once you sit with it for a while.
Looking At The Bigger Scaling Picture
Ethereum has spent years reducing the cost of individual operations while raising the ceiling on what a single block can contain. Batch proofs fit naturally into that trajectory. Instead of every statement carrying its own full verification cost, related statements can share the expensive parts. The draft simply adds a practical payment rail on top of that technical capability.
Other networks have experimented with similar ideas under different names. What feels distinctive here is the explicit sponsor-member split and the insistence on keeping individual signatures and nonces. That combination may prove more user-friendly than systems that force everyone into a single collective signature.
Open Questions That Still Need Answers
Several practical questions remain. How often will sponsors step forward in live conditions? Will the provisional gas numbers shrink once real hardware measurements arrive? Can the dependency limits stay high enough to be useful without creating new denial-of-service vectors? The draft itself flags the need for full-block performance data before any activation decision.
I’ve found that the healthiest proposals are the ones that list their own uncertainties clearly. This one does. It does not pretend the fee schedule is final. It does not claim activation is around the corner. It simply places a concrete design on the table and invites measurement.
- Sponsor covers the shared leaf proof
- Members pay only for their individual claims
- Builders can include any valid subset without new signatures
- Nonces and receipts stay attached to each transaction
- Provisional gas figures still require calibration
Why The Timing Feels Relevant
Layer-two activity continues to grow, yet base-layer efficiency still matters. Every reduction in the cost of verifying complex statements helps the whole stack. If sponsored batches become reliable, applications that currently avoid heavy cryptography for cost reasons might reconsider. That secondary effect could prove larger than the direct gas savings.
At the same time the draft stays carefully scoped. It introduces no new networking messages. Commitment details remain in the existing proof-system specification. The change surface is smaller than many previous upgrades, which may help the review process move more smoothly once formal discussion begins.
A Personal Take On The Direction
In my experience the proposals that age best are the ones that respect existing mental models. Users already understand that they pay for their own transactions. Sponsors already understand that they can pay for services. This draft simply lets those two existing behaviors meet in a new place. It does not ask anyone to learn an entirely new economic language.
That restraint is welcome. Crypto has seen its share of clever designs that required users to rethink everything at once. The ones that stick tend to feel like natural extensions of what people already do. Shared batch proofs with separate payments strike me as closer to the second category.
Next Steps The Community Will Watch
An EIP number still needs assignment. A dedicated discussion thread will almost certainly open. Performance measurements covering full blocks, late repairs, and realistic gas overhead must be recorded. Only after those pieces exist can anyone seriously discuss activation timelines.
Until then the draft remains exactly what it claims to be: a working proposal under active refinement. The core idea of separating the proof work from the individual payment work looks solid. Whether the concrete numbers deliver the expected savings will become clearer once the benchmarks arrive.
I’ve followed enough Ethereum upgrades to know that the interesting phase often begins after the first draft. Comments arrive, edge cases surface, and the numbers get stress-tested. This particular draft starts from a place that already feels thoughtful. That gives me cautious optimism about where the conversation might lead.
Practical Implications For Different Participants
For ordinary users the change could remain mostly invisible if sponsors become common. Transactions simply become cheaper when they ride inside a sponsored batch. For power users and application developers the new payment rails open design space. An application could choose to sponsor its own users’ proofs as a feature, or third-party sponsors could compete on efficiency.
Builders gain a new revenue stream from sponsor payments and discounted claims. They also inherit a new risk when a sponsorship later fails validation. How that risk is priced into inclusion decisions will shape the practical economics more than any single gas figure.
Keeping Expectations Realistic
It is easy to read an early draft and project finished results. I try to resist that urge. The current gas numbers are placeholders. The resource limits are starting points. The performance requirements are explicitly listed as unfinished business. Treating the draft as a finished upgrade would be a mistake.
At the same time dismissing the idea because the numbers are still provisional would also miss the point. The architectural choice to keep payments and signatures separate while sharing the proof work is the real contribution. Everything else can be measured and adjusted.
Where The Conversation Goes From Here
The draft is public. Reviewers will examine the interaction between recursive proofs and payment finalization. Others will pressure-test the dependency limits. Still others will run the full-block measurements the proposal itself demands. Each of those steps will refine the design further.
For anyone who follows Ethereum’s technical evolution, this is the kind of proposal worth tracking. It does not promise overnight transformation. It does offer a concrete mechanism that could make batch verification more economically practical for a wider range of users. In a network that continues to balance decentralization with efficiency, that kind of practical progress still matters.
The coming weeks and months will show whether the provisional figures hold up and whether sponsors appear in sufficient numbers. Until those answers arrive, the draft stands as a thoughtful attempt to solve a real friction point. I’ve seen less promising ideas move further, and I’ve seen stronger-looking ideas stall. The only reliable path is the one the draft itself recommends: measure thoroughly, then decide.
That measured approach is itself a quiet strength. In a space that sometimes rushes, a proposal that insists on full-block evidence before activation feels refreshingly grounded. Whether the final numbers deliver the hoped-for savings or require further adjustment, the conversation has already become more precise than it was a week ago. And that, in my view, is already useful progress.
