hegota eip selection explainer

Hegotá EIP Selection — Explainer

This page walks through our team's tier ranking of the execution-layer (EL) EIPs proposed for Hegotá. It's meant as a reference for anyone who wasn't in the room when we ranked these, not a line-by-line rationale for all 37 EIPs.

Our full ranking is viewable on Forkcast

Thematically:

We lean hard into 8141 and its extensions, and cleanup techdebt.

S tier — must ship

EIP	Name	Why it's here
5920	PAY Opcode	Small, low-surface addition that closes a long-standing gap: sending value without invoking recipient code.
7668	Remove Bloom Filters	Cleans up a receipt/logs mechanism that's largely obsolete now that clients index logs differently.
7906	Part of the Frames extension package alongside EIP-8298 and EIP-8151. We think frames extensions make sense to ship for immediate use with EIP-8141
8250	Keyed Nonces for Frame Transactions	Frames-core support.
8253	Bump Nonce of Zero-Nonce Storage Accounts	State-transition cleanup that meaningfully simplifies the upcoming trie migration. Pays down techdebt
8368	CPSB Recalibration for New Gas Limit	Recalibrates cumulative pricing/state-bloat controls for the new gas limit set by Glamsterdam.

A tier — high priority, please ship

EIP	Name	Why it's here
8355	ML-DSA Verification Precompiles	Post-quantum work needs to start ASAP, but 3 precompiles might be a bit much.
3298	Removal of Refunds	Deletes an entire class of gas-refund metering edge cases the last fork paid for dearly.
4758	Deactivate SELFDESTRUCT	Removes the last live SELFDESTRUCT path — a testing hazard every future EIP otherwise has to define behavior against. Techdebt pay down.
7666	EVM-ify the Identity Precompile	Rewrites the identity precompile as EVM bytecode instead of native client code — a proof point for EVMification (EIP-8200). We like this one because it is simple, and a useful proof of concept.
7862	Delayed State Root	Easy to implement, performance improvement.
8131	Unified Transaction Content Floor	Bounds worst-case calldata pricing; pairs with EIP-8279 as the other half of the bounding work.
8151	Account Code Restricted ecRecover	Valuable Post-Quantum work
8279	Block Access List Byte Floor	Floors adversarial Block Access List pricing so worst-case block construction stays bounded.
7709	Read BLOCKHASH from Storage and Update Cost	zkVM witnesses benefit from this, reducing their size significantly

B tier — on the bubble

EIP	Name	Why it's here
7979	Call and Return Opcodes for the EVM	Adds new control-flow opcodes; reasonable, but adds surface to a fork already carrying two headliners worth of interaction testing.
8077	eth/XX: Announce Transactions with Nonce	Richer mempool announcements for the Frames era; needs sophisticated network simulation.
8163	Reserve EXTENSION (0xae) Opcode	A reservation with no committed consumer yet — can ride a later fork.
8272	Recent Roots for Frame Transactions	Frames-core: lets private transactions reference recent on-chain state in an attester-verifiable form.
8374	Persist Warm Access Sets Across Reverts	Net-metering repricing; paired with EIP-8358 and shouldn't carry a different grade than it does.

C tier — not disqualified, but ehhhhhhhh

EIP	Name	Why it's here
2488	Deprecate the CALLCODE Opcode	Low urgency.
7807	SSZ Execution Blocks	Wide-blast-radius formatting migration with no Hegotá-specific dependency forcing it now.
7819	SETDELEGATE Instruction
7851	Code-Controlled EOA Delegation
7923	Linear, Page-Based Memory Costing	Repricing-family change; it's good, we like it, not sure the value yield is there
8115	Batch Priority Fees at End of Block	Same repricing-churn concern as 7923.
8116	Replace Cumulative Receipt Fields	performance yield is low, but it's good tech-debt cleanup.
8188	Last-Written Block for Accounts and Slots
8200	EVMification	We like EVMification as a direction, but flagged this one as feeling premature — every replacement bytecode needs an exact-behavior audit, is it fast enough at runtime?
8298	SETCODEFROM Code Reuse Instruction
8372	Normalized State Gas Limit

D tier — declined for inclusion

EIP	Name	Why it's here
7645	Alias ORIGIN to SENDER	Breaks assumptions baked into deployed contracts for no offsetting security gain.
8094	eth/vhash: Blob-Aware Mempool	we questioned why this needs a fork at all if a sparse blobpool already captures most of the benefit.
8182	Private ETH and ERC-20 Transfers	Enshrines one specific privacy mechanism at the protocol layer; the Frames-based path gets similar outcomes with less protocol surface and keeps schemes competing.
8219	Checked Arithmetic Opcodes	Good idea, compiler teams can already pursue, not urgent
8304	Trustless Log and Transaction Index
8358	Net Gas Metering for Account Changes

Open questions to close out

A few items above still need an answer before we'd call this ranking final:

• EIP-2488: has anyone actually verified the deprecation justification, or is it still an unvalidated claim?
• EIP-8200 / EIP-7666 (EVMification): we're supportive of the direction, but what stops the EVM-ified precompiles from just getting optimized back into native precompiles later.
• EIP-8094: the bandwidth/mempool savings case hasn't been made convincingly enough to justify a fork-scoped change, especially with sparse blobpool already helping here.