EIP-7923: Linear, Page-Based Memory Costing: When Ethereum Fixes One Problem and Creates Another

The proposal appeared on the Ethereum Magicians forum in September 2026 with the technical precision of an engineer who had spent months profiling gas costs and was ready to fix what they saw as an obvious inefficiency. EIP-7923 proposed replacing Ethereum's quadratic memory expansion cost model with a linear, page-based alternative. The pitch was optimization: quadratic costs unfairly penalized memory-intensive operations, made certain contract patterns prohibitively expensive, and created unpredictable gas spikes for legitimate use cases. The fine print raised a question about whether a change that made memory cheaper would also make memory abuse cheaper, and whether the contracts that were supposed to benefit from lower costs would find themselves competing with a new class of gas-guzzling operations that could previously never afford to run.

That was the proposal. Then came the question of whether the Ethereum protocol that was supposed to charge for resource consumption in proportion to network load was now planning to discount the resource that had been deliberately expensive because it was the easiest one to abuse.

What EIP-7923 Actually Proposes

The EIP describes a fundamental change to EVM memory pricing with specific mechanics:

The Core Components:

- Page-based memory: Memory is allocated in 256-byte pages rather than byte-by-byte

- Linear expansion cost: Each additional page costs a fixed amount instead of the current quadratic formula

- Cheaper memory operations: MLOAD and MSTORE operations become less expensive at higher memory sizes

- Simplified gas accounting: The quadratic memory_word_gas formula is replaced with linear page pricing

The Current System:

- Quadratic cost: Memory expansion gas increases with the square of memory size

- Byte granularity: Memory is charged per 32-byte word with progressively higher costs

- Abuse protection: Quadratic costs naturally limit memory-heavy operations

- Predictable pain: Developers know that high memory usage costs exponentially more

The Proposed Changes:

- Flattened costs: Memory expansion becomes cheaper at higher sizes

- Page granularity: Memory is allocated in fixed-size chunks

- Reduced gas for large memory: Contracts using substantial memory pay proportionally less

- Encouraged memory usage: Operations that were previously too expensive become viable

The EIP frames these as efficiency improvements. They are also a fundamental change to Ethereum's resource pricing philosophy.

Key Metrics at a Glance

The Proprietary Memory Pricing Integrity Score (MPIS)

I've developed a framework to evaluate whether linear memory pricing preserves or undermines Ethereum's resource economics:

Formula: MPIS = (Abuse Resistance × 0.3) + (Resource Honesty × 0.25) + (Network Stability × 0.2) + (Developer Fairness × 0.15) + (State Sustainability × 0.1)

EIP-7923 Assessment:

A score of 4.2 indicates that EIP-7923 is a mixed-efficiency proposal: genuine benefits for developers but substantial risks for network security and validator economics.

The Three Memory Pricing Traps

Trap 1: The DoS Discount

The fundamental problem with linear memory pricing is that it removes a DoS protection that was working. The quadratic cost model was not an accident or a bug—it was a deliberate design choice that made memory exhaustion attacks prohibitively expensive. EIP-7923 replaces that protection with a cheaper alternative. The attacker who previously could not afford to fill memory now can. The contract that previously could not be exploited through memory abuse now can be. The protocol that was supposed to be secure by economics is now secure by hoping no one attacks it. The discount that was supposed to help developers also helps attackers.

Trap 2: The Hidden Node Cost

Memory is not free for nodes. When a contract uses 1MB of memory during execution, that memory must be allocated, managed, and freed. The node operator pays the cost in RAM, in validation time, and in hardware requirements. The quadratic pricing model compensated nodes for that cost by making large memory expensive. The linear model does not. The validator that was supposed to be incentivized to participate finds their costs increasing while their rewards stay flat. The decentralization that was supposed to be preserved by accessible validation is undermined by memory-intensive blocks that require more hardware.

Trap 3: The State Growth Acceleration

Cheaper memory means more memory usage means larger contract state means faster state growth. The state that Ethereum validators must keep in memory or SSD grows with every contract that decides to use more memory. The pruning, archiving, and sync processes that keep the network manageable become more expensive. The node that was supposed to sync in hours now syncs in days. The validator that was supposed to run on a modest server now needs enterprise hardware. The long-term sustainability that was supposed to be preserved by economics is accelerated by a pricing change.

Competitive Landscape: Memory Pricing Across Blockchains

The landscape shows that linear memory models exist but come with trade-offs. Arbitrum Stylus's WASM model is the closest parallel and already shows signs of memory-intensive operations becoming common.

Scenario Analysis: Three Futures for EVM Memory Economics

Scenario A: Memory Abuse Normalization (40% probability)

• Contracts begin using substantially more memory than before
• Block gas limits are hit more frequently by memory-heavy transactions
• Validators require more RAM to keep up
• MPIS degrades to 2.5/10

Scenario B: Targeted Optimization (35% probability)

• The change benefits specific use cases (large data processing, batch operations)
• Most contracts continue using similar memory patterns
• The DoS surface expands modestly
• MPIS stabilizes at 5.0/10

Scenario C: Partial Reversion (25% probability)

• Memory abuse incidents prompt emergency adjustments
• A hybrid model is adopted with linear costs up to a threshold and quadratic beyond
• The EIP is partially rolled back
• MPIS improves to 5.5/10

The Bottom Line

EIP-7923 is technically sound engineering from a team that understands gas costs and wants to improve developer experience. The Memory Pricing Integrity Score is 4.2/10. Abuse resistance is weakened. Resource honesty is compromised. Network stability is at risk. Developer fairness is improved. State sustainability is challenged.

The three traps—DoS discount, hidden node cost, and state growth acceleration—are structural risks that accompany every attempt to make memory cheaper on a blockchain. They reflect the fundamental tension between developer convenience and network security in protocol design. The community that wants cheaper memory must also accept that cheaper memory is cheaper for everyone, including those who want to harm the network.

The deeper question is whether Ethereum can afford to discount its primary DoS protection. The entire value proposition of Ethereum's gas model is that resource consumption costs what it costs. The protocol that subsidizes memory is a protocol that has abandoned honest pricing. The developer who benefits from lower memory costs is subsidized by the validator who pays for it through higher hardware requirements.

The EIP deserves recognition for identifying a real problem. Quadratic costs do create perverse incentives and limit legitimate use cases. But the solution that helps developers while hurting validators is not a net improvement. The protocol that optimizes for developer convenience over network security has optimized for the wrong thing. The memory that becomes cheaper for everyone becomes a weapon for those who would attack the network.

TL;DR

• What: EIP-7923 proposes replacing Ethereum's quadratic memory expansion cost model with a linear, page-based alternative, making large memory operations substantially cheaper
• The Score: Memory Pricing Integrity Score of 4.2/10—abuse resistance (3/10) weakens as memory attacks become affordable; resource honesty (4/10) fails to compensate nodes for actual memory costs; network stability (4/10) degrades as validators need more hardware; developer fairness (7/10) improves for legitimate use cases; state sustainability (4/10) accelerates growth with cheaper memory incentives
• The Reality: A genuine efficiency improvement for developers that transfers costs to validators and expands the DoS surface
• Three Traps: DoS discount (memory attacks become affordable); hidden node cost (validators pay for cheaper developer memory); state growth acceleration (larger contract state grows faster)
• Outlook: Memory abuse normalization (40%) where contracts use substantially more memory; targeted optimization (35%) where specific use cases benefit without broad abuse; partial reversion (25%) where incidents prompt emergency adjustments

Sources

• Ethereum Magicians - EIP-7923 Discussion - September 2026 proposal for linear page-based memory costing
• EVM Memory Documentation - Current memory expansion gas mechanics and formulas
• Yellow Paper - Memory Cost Formula - Formal specification of quadratic memory pricing
• Arbitrum Stylus Documentation - WASM linear memory model and observed usage patterns
• Ethereum Research - State Growth Analysis - Academic discussion of state size and pruning challenges
• Gas Price Oracle Data - Historical gas costs for memory-intensive operations
• The Block - EVM Optimization Proposals - Analysis of protocol efficiency improvements and trade-offs
• Client Developer Discussions - Validator perspectives on memory requirements and hardware costs

Zain Tran is TotesTek's Ethereum Ecosystem Columnist & Accountability Reporter. He writes about Ethereum, ETH, smart contracts, DeFi, Layer 2 networks, staking, validators, and the real-world consequences of technical and financial failure.