Mempools Across Chains

When a node receives a valid transaction that is not yet confirmed, it can store that transaction in a memory pool. Each node makes this decision locally according to protocol rules and its own policy. Two honest nodes can therefore have different mempools while agreeing perfectly on the confirmed blockchain.

That distinction matters because wallets and explorers often display “pending” as if there were one global queue. In reality, pending status is an observation from particular network participants.

A Bitcoin node validates a new transaction against its current UTXO set and local policy before adding it to the mempool. The transaction is announced to peers, which independently make the same decision. Nodes can reject low-fee transactions, evict entries when memory limits are reached, or remove transactions that conflict with newly confirmed blocks.

Miners choose candidate transactions from their own mempool according to fee and policy preferences; there is no protocol rule requiring every miner to use the same queue.

Ethereum execution clients also maintain local pools of signed transactions, commonly organized by sender nonce and fee conditions. Searchers can observe public pending order flow and compete for ordering opportunities. Other networks use leader schedules, private relays, or different forwarding paths that make the public pending set less globally visible or structurally different.

Calling all of these systems “the mempool” can hide major differences in who sees transactions before inclusion and who can influence ordering.

When pending demand exceeds block capacity, block producers have more transactions to choose from than they can include. Fee rates become a sorting signal. Transactions paying too little can wait, be evicted, or require replacement with a higher-fee version where protocol rules allow.

Fee estimation therefore depends on current mempool composition, producer behavior, and expected future demand rather than a fixed network price.

Public mempools reveal transaction intent before execution. Arbitrageurs can react to swaps, attackers can attempt front-running or sandwich strategies, and observers can correlate transaction timing with network sources. Private submission channels can reduce some exposure but introduce new trust assumptions.

Mempool architecture therefore influences both market microstructure and privacy.

A mempool is local state maintained by a node, not a consensus object shared identically by the entire network. When a node receives a valid but unconfirmed transaction, it may store it temporarily and relay it. Another node can have a different mempool because it learned about different transactions, applies different fee thresholds, has different storage limits, or evicted older entries.

This matters for user expectations. Broadcasting a transaction does not guarantee that every block producer has seen it. A transaction can disappear from one mempool because of replacement, expiry, conflict, fee policy, restart, or chain-specific rules while still existing elsewhere.

Different blockchains use very different pending-transaction architectures. Some have conventional public mempools. Others use private order flow, leader-specific queues, local forwarding, or specialized transaction pipelines. Even within one ecosystem, builders and validators may receive transactions through channels not visible in a public node's mempool.

Mempool policy also influences transaction ordering and fees. Nodes or block producers can prioritize by fee, age, dependency, local policy, or application-specific rules. Because this policy is often outside consensus, two honest nodes can make different choices without disagreeing about chain validity.

The safest mental model is “candidate transactions known to this node right now,” not “the blockchain's official waiting room.”

1. Why can two honest nodes have different mempool contents?

2. Is a mempool part of consensus state? Explain.

3. List three reasons a transaction can disappear from one node's mempool.

4. Why does transaction broadcast not guarantee block inclusion?

5. How can private order flow make a public mempool an incomplete picture of pending demand?

6. Correct this statement: “There is one global mempool shared by every blockchain node.”