Block Template Construction: The Hidden Centralization Risk in Bitcoin Mining

Bitcoin's security model assumes that no single entity controls which transactions get included in blocks. In practice, a handful of mining pools construct the block templates that determine exactly which transactions enter the blockchain. As of mid-2026, just three pools control over 61% of hashrate, and the operators of those pools decide what goes into every block they mine. Individual miners who contribute hash power have no say in which transactions are included or excluded.

This is the block template centralization problem: the gap between who performs proof-of-work computation and who decides what that computation commits to. It creates a censorship vector that most Bitcoin users never think about, and it has already been exploited in practice.

How Block Template Construction Works

A block template is the blueprint for a Bitcoin block. It specifies which transactions from the mempool to include, in what order, and how to structure the coinbase transaction that pays the miner. The template also includes the previous block hash, timestamp, and difficulty target. Once constructed, miners hash against this template until one finds a valid proof of work.

The getblocktemplate era

In Bitcoin's early years, miners constructed their own templates. The getblocktemplate RPC (defined in BIP 22) let each miner query their local Bitcoin Core node for a candidate template, modify it freely, and submit completed blocks. Every miner chose which transactions to include based on their own fee-rate preferences and policy rules. Template construction was fully decentralized.

Stratum V1 changed everything

When Stratum V1 launched in 2012 (created by Slush Pool, now Braiins Pool), it replaced getblocktemplate with a lightweight JSON-RPC protocol optimized for pooled mining. The pool server constructs the block template internally and sends miners only the minimal data needed to produce valid block headers: the coinbase transaction, Merkle branch hashes, and target. This is roughly 1 kilobyte of data per job assignment.

The tradeoff was stark: Stratum V1 made pooled mining practical at scale by reducing bandwidth and latency, but miners lost all autonomy over block construction. Individual miners now hash blindly against whatever template the pool provides. They cannot see the full transaction list, cannot add transactions, and cannot remove them.

The core problem: Under Stratum V1, template construction is entirely controlled by pool operators. Miners contribute hash power but have zero influence over which transactions enter the blockchain. The entity performing the work and the entity deciding what the work commits to are completely separated.

How Concentrated Is Template Construction Today?

The degree of concentration is worse than most people realize. According to Hashrate Index data for mid-2026, the Bitcoin mining landscape is dominated by a small number of pools, each of which independently decides which transactions to include in their blocks.

The Nakamoto coefficient for block template construction is just three: only three pool operators need to coordinate to control over 50% of all template decisions. Research from Bitcoin developer 0xB10C's Mining Pool Observer has shown that AntPool and its associated proxy pools controlled approximately 40% of network hashrate in early 2025, meaning that when combined with Foundry at roughly 30%, two entities directed block template construction for about 70% of all Bitcoin blocks.

The total network hashrate peaked above 1 ZH/s (1,000 EH/s) in late 2025 before declining to approximately 866 EH/s by mid-2026, partly due to competition from AI data centers for the same electrical infrastructure that Bitcoin miners rely on. But the concentration ratios have remained roughly stable: the top pools have maintained their relative share even as total hashrate fluctuated.

When Pools Censor Transactions

Block template centralization is not a theoretical concern. Multiple pools have been caught filtering transactions based on criteria unrelated to fee rates, directly undermining Bitcoin's censorship resistance guarantees.

Marathon Digital: OFAC-compliant mining (2021)

In May 2021, Marathon Digital launched what it called an "OFAC-compliant" mining pool that excluded transactions from addresses on the U.S. Treasury's Specially Designated Nationals (SDN) list. Block 682,170, mined on May 5, 2021, contained only 178 transactions and earned $2,903 in fees, while adjacent blocks processed over 1,000 transactions each and earned approximately $17,500 in fees. The pool was voluntarily leaving money on the table to comply with sanctions.

Following intense community backlash, Marathon CEO Fred Thiel announced on May 31, 2021 that the pool would stop censoring transactions. But the precedent had been set: a major public mining company had demonstrated willingness to filter transactions at the protocol level.

F2Pool: OFAC filtering (2023-2025)

In late 2023, 0xB10C's miningpool-observer detected six Bitcoin transactions spending from OFAC-sanctioned addresses that were missing from blocks. Four of the six blocks were mined by F2Pool. Co-founder Chun Wang initially posted on X justifying the filtering before deleting the post. F2Pool publicly admitted to running a "tx filtering patch" and announced it would be disabled "until the community reaches a more comprehensive consensus."

But the filtering returned. In late 2024 through January 2025, 0xB10C detected fifteen additional missing OFAC-sanctioned transactions. Of 14 analyzed blocks, 11 were mined by F2Pool, with 9 showing patterns consistent with intentional filtering. F2Pool's most recent inclusion of an OFAC-sanctioned transaction was over seven months prior, suggesting the filtering patch had been quietly re-enabled.

Why this matters: All filtered transactions eventually confirmed in subsequent blocks mined by other pools. With multiple pools operating, single-pool censorship creates delays rather than permanent exclusion. But if multiple large pools adopted the same filtering criteria, transactions could face indefinite delays or effective exclusion from the blockchain entirely.

Private mempools and MEV extraction

Beyond sanctions compliance, pools have begun offering private transaction submission services that bypass the public mempool. In February 2024, Marathon Digital launched "Slipstream," a service where users can submit large or non-standard transactions directly to Marathon's mempool. While pitched as a convenience feature, private mempools create an information asymmetry: pool operators see transactions that other participants cannot, enabling MEV-like extraction strategies.

Bitcoin's MEV landscape differs from Ethereum's (no complex AMMs or smart contract interactions on L1), but subtler forms are emerging: inscription sniping via PSBT front-running on Ordinals marketplaces, fee spike exploitation during protocol launches like Runes, and the cartelization risk inherent in pooled mining where consecutive block production enables multi-block extraction strategies.

Why Individual Miners Cannot Resist

Under Stratum V1, an individual miner who disagrees with their pool's template policy has exactly one option: leave the pool. They cannot construct their own template while remaining in the pool, and the economics of solo mining are prohibitive. With network hashrate at approximately 866 EH/s, even a large mining operation with 10 EH/s of capacity would find a block roughly once every 10 days on average. The revenue variance is unacceptable for any business with fixed costs for electricity, facilities, and equipment.

This creates a structural dependency: miners need pools for revenue smoothing, but joining a pool means surrendering template construction authority. The pool operator becomes a single point of control over which transactions the miner's hash power commits to.

Stratum V2: Returning Template Construction to Miners

Stratum V2 is the primary protocol-level solution to template centralization. Developed initially by Braiins (the successor to Slush Pool, which created Stratum V1), the Stratum V2 specification introduces three sub-protocols that collectively modernize the mining communication stack.

The three sub-protocols

The Mining Protocol is the direct successor to Stratum V1, handling communication between mining devices and pools. It uses binary framing (reducing bandwidth by approximately 70% compared to V1's JSON-RPC) and mandates Noise Protocol encryption, eliminating the man-in-the-middle attacks possible with V1's unencrypted connections.

The Template Distribution Protocol replaces getblocktemplate with a push-based mechanism for receiving block template data from Bitcoin Core, reducing latency when new blocks are found.

The Job Declaration Protocol is the critical innovation for decentralization. It allows a miner to run their own Bitcoin Core node, construct a block template from their local mempool based on their own criteria, and propose that template to the pool. The pool validates the template (ensuring correct work attribution and share accounting) and, if approved, the miner hashes against their own template rather than the pool's. A single job declaration can be reused across an entire mining farm with hundreds of thousands of devices.

Adoption status

As of mid-2026, two pools run Stratum V2 natively in production: Braiins Pool and DEMAND Pool (DMND, launched November 2025 as the first V2-native pool). On May 7-11, 2026, seven major pools representing approximately 75% of global Bitcoin hashrate joined the Stratum V2 Working Group: Foundry USA, AntPool, F2Pool, SpiderPool, MARA Pool, Block Inc., and DMND. These pools are in testing and integration phases.

A landmark milestone occurred on June 25, 2026, when GoMining and DMND pool mined block 955,318 using the Job Declaration feature. This was the first known Bitcoin block where a miner constructed its own template in a live Stratum V2 production environment. GoMining included transactions from GoBTC Pay, its own open-source Bitcoin payments protocol, demonstrating the kind of template sovereignty that Job Declaration enables.

OCEAN and the DATUM Protocol

OCEAN pool, founded by Bitcoin Core developer Luke Dashjr and backed by a $6.2 million seed round led by Jack Dorsey, takes a different approach to decentralized template construction. Launched in November 2023 with approximately 2% of global hashrate, OCEAN developed DATUM (Decentralized Alternative Templates for Universal Mining), a protocol built from scratch for miner-constructed templates.

Under DATUM, miners run a DATUM gateway alongside a Bitcoin full node on the same device. The gateway communicates with mining hardware and constructs templates from the miner's local mempool. Coinbase payouts go directly to miners (non-custodial), and miners using DATUM receive a 50% discount on pool fees as an incentive.

OCEAN distinguishes DATUM from Stratum V2 by arguing that V2's Job Declaration is "decentralization bolted onto the original centralized design," while DATUM was purpose-built for decentralized template construction. In April 2025, Tether announced it would deploy hashrate on OCEAN, lending additional visibility to the DATUM approach.

Comparing the Mining Protocol Approaches

Three distinct approaches now compete for how Bitcoin block templates should be constructed, each with different tradeoffs for decentralization, efficiency, and adoption.

The Broader Decentralization Stakes

Block template centralization affects more than just which L1 transactions get confirmed. It has cascading implications for every system that depends on Bitcoin's censorship resistance as a settlement layer.

L2 protocols depend on uncensorable L1 access

Layer 2 protocols like Spark, Lightning, and statechains all rely on the ability to broadcast transactions to Bitcoin L1 when needed. Lightning channels require timely justice transactions to punish cheating counterparties. Spark users need the ability to execute unilateral exits to L1 if operators become unavailable. If mining pools can selectively delay or exclude these time-sensitive transactions, the security guarantees of every Bitcoin L2 are weakened.

This is not hypothetical: Lightning's penalty mechanism relies on commitment transactions reaching the blockchain within specific timelock windows. A pool that censors a justice transaction for long enough could enable a channel counterparty to steal funds. Similarly, Spark's exit transactions use relative timelocks to establish ownership hierarchy. Delayed inclusion would not prevent the exit entirely, but it could create windows where competing claims overlap.

The regulatory pressure vector

The documented censorship cases all involve OFAC sanctions compliance, a regulatory framework that applies to U.S.-based entities. Foundry USA, the largest pool by hashrate, is operated by Digital Currency Group, a U.S. company. MARA Pool is a subsidiary of Marathon Digital Holdings, a Nasdaq-listed company. Both are subject to U.S. regulatory pressure in ways that non-U.S. pools are not.

If template construction remains concentrated at the pool level, a regulatory order requiring transaction filtering by U.S.-based pools could affect over 35% of all Bitcoin blocks. Combined with voluntary compliance by pools in allied jurisdictions, the impact could be even larger. Decentralized template construction neutralizes this vector: when individual miners build their own templates, no single regulatory jurisdiction can dictate inclusion policy across a meaningful share of hashrate.

The path forward

The May 2026 announcement that pools representing 75% of hashrate have joined the Stratum V2 Working Group is significant, but joining a working group is not the same as deploying Job Declaration in production. The Stratum Reference Implementation (SRI) working group projects that V2 could become the default protocol for new ASIC firmware shipments by end of 2026, potentially reaching 40-60% of network hashrate. BraiinsOS+ remains the only commercially available firmware with a native Stratum V2 client, supporting Antminer S9, S19, S21, and S21 XP series.

The practical barrier to adoption is not protocol design but infrastructure cost. Job Declaration requires each mining operation to run a Bitcoin Core full node, maintain mempool synchronization, and manage the additional complexity of template construction alongside ASIC operations. For large institutional miners with dedicated engineering teams, this is straightforward. For smaller operators, it adds operational burden. OCEAN's DATUM protocol tries to lower this barrier by bundling the node and gateway on a single device, but the fundamental requirement for each template-constructing miner to maintain their own view of the mempool remains.

What Developers and Users Should Watch

Several metrics indicate whether template decentralization is actually progressing or merely being discussed.

• The percentage of blocks mined with miner-constructed templates (currently under 3%) is the most direct measure. Tools like miningpool.observer and mempool.space track pool-level block production.
• ASIC firmware adoption of Stratum V2 is a leading indicator. If major manufacturers like Bitmain and MicroBT ship V2 clients as the default, the migration accelerates.
• Template diversity across pools can be measured by comparing the transaction sets in blocks from different pools. Identical templates across pools suggest shared infrastructure; divergent templates suggest independent construction.
• Incidents detected by 0xB10C's observer tooling continue to be the primary accountability mechanism for pool-level censorship.

Implications for Spark and Bitcoin L2s

For protocols like Spark that depend on censorship-resistant L1 settlement, block template decentralization is a foundational concern. Spark's security model ensures that users can always exit to Bitcoin L1 via pre-signed exit transactions, even if all operators become unavailable. That guarantee is only as strong as the L1's willingness to include those transactions without discrimination.

The progress on Stratum V2 and DATUM represents meaningful movement toward reducing this risk. As more miners construct their own templates, the attack surface for coordinated censorship shrinks. Developers building on Bitcoin L2s should track the adoption of miner-constructed templates as a key indicator of the settlement layer's health. For those building on Spark, the Spark SDK documentation covers how unilateral exits interact with L1 fee market dynamics and block space availability.

For a deeper look at how mining economics interact with protocol incentives, see our coverage of Bitcoin mining economics in 2026 and Stratum V2's role in mining decentralization.

This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.