Network Upgrades
Three Ways Crypto Addresses Carry Network Context
Prefixes, checksums and chain-aware wrappers reduce wrong-network transfers, but only explicit identifiers resolve the ambiguity shared by EVM layer 2s.
ERC-7930, created on February 2, 2025, makes the three ways crypto addresses carry network information unusually clear: a native prefix, a chain-bound checksum, or an explicit envelope around account bytes. For a layer-2 operator, the distinction matters at gateways and indexers, not merely in wallet design. A parser must decide which chain an account belongs to before the sequencer accepts a transaction or a bridge routes an asset.
Why does 0x reveal so little?
An EVM address is a 20-byte account identifier; the “0x” only marks hexadecimal notation, not Ethereum, Base or an optimistic rollup. The same private key can therefore produce the same account address across EVM networks, while balances and contract code differ on each chain.
Network separation normally arrives elsewhere. The wallet signs a transaction containing a chain ID and submits it through a chain-specific RPC endpoint. The sequencer orders it, L2 nodes re-execute it against that rollup’s state, and batch data is ultimately retained through the rollup’s chosen settlement and data-availability path. A correct address sent through the wrong endpoint can still target the wrong state.
1. Native prefixes identify the decoder
Some address formats put network context directly in their printable form. Bech32 uses a human-readable prefix before its separator; other schemes place a network or address-type byte inside a checksummed payload. A wallet decodes that field, verifies the checksum and rejects a prefix it does not support.
This is operationally cheap because the prefix need not enter consensus data: the transaction carries the decoded destination. The failure point is the software registry mapping prefixes to networks. EVM rollups usually preserve the 20-byte hex form for compatibility, so adopting a new native prefix would impose wallet, RPC and tooling changes without reducing execution or data-availability costs.
2. Can capitalization bind an address to a chain?
Yes, but only when every participant enforces the same checksum rule. ERC-1191 folds the chain ID into the hash that determines which hexadecimal letters are uppercase, creating a chain-aware presentation while leaving the underlying 20 bytes unchanged. A wallet can catch a mismatch before submission; a parser that lowercases the string discards the protection.
The Manta Bridge view is a useful editorial reference for the operational setting: a bridge must pair an account with a route and token policy even when the displayed hex address carries no network label. Checksummed case adds no calldata or storage fee after decoding, but inconsistent adoption makes it a warning layer rather than a universal boundary.
3. Explicit wrappers move context into data
Explicit formats bind a chain identifier to the account rather than asking software to infer it. CAIP-10 does this as a text string; ERC-7930 proposes a versioned binary envelope containing a chain type, chain-reference length, chain reference, address length and address. The user or wallet submits the pair, a gateway decodes it, and a contract can verify or store it when the context must survive on-chain.
- Native prefix: readable and cheap, but tied to format-specific registries.
- Chain-bound checksum: backward-friendly and fee-neutral, but fragile when software normalizes case.
- Explicit envelope: unambiguous and extensible, but costs more bytes and parsing work when persisted on-chain.
The operator verdict
Explicit chain-and-account identifiers are the practical improvement for a multi-rollup stack because validation can happen once at the ingress boundary and the result can follow a message across services. They do not make sequencers more decentralized, batches cheaper or data more available. ERC-7930 remains under review, and no measured production result yet proves that its extra structure reduces failed transfers or support load at scale. Until wallets, bridges and RPC providers converge on one encoding, operators should preserve chain ID as a separate validated field and reject ambiguous routes.
Filed under
- Network Upgrades
- Sequencer Operations