Imagine watching a Bitcoin payment accumulate confirmations, then discovering that your wallet has been counting blocks a fully validating node will never accept.
That was the danger during Bitcoin’s validationless-mining incidents in July 2015. Some miners kept finding blocks and spending electricity. Their machines were working. The chain they were extending was invalid.
The lesson was expensive, and beautifully simple: proof of work cannot turn a broken rule into a valid transaction.
Now a researcher is asking whether mining can deliberately run ahead of full validation—and whether the protocol can contain the consequences when something goes wrong.
On September 22, 2026, ** Gyu Chol Kim presented a framework called Asynchronous Pipeline Consensus on Delving Bitcoin**. The proposal separates the information needed to begin the next mining round from the full transaction data that nodes must eventually verify.
That is an attention-grabbing number. It is also the least interesting part of the story.
The deeper question is whether Bitcoin could change the timing of verification while preserving the authority of every independent node to reject invalid money.
If the answer were yes, it could open a different route through the scaling debate. If the answer is no, the reason may teach us more than another impressive throughput chart ever could.
To see why, we need to return to the mistake that made this idea sound dangerous in the first place.
Oil above $100. Treasury yields above 5%. A Fed hike. Altman, Amodei, and Huang are helping buy America time—and AI has to justify the bet.
In July 2015, Bitcoin was enforcing ** the BIP66 upgrade**, which introduced strict signature-encoding requirements. Its activation rules also required upgraded participants to reject old version-2 blocks after the threshold was reached. An outdated miner produced such a block. Other miners extended it without doing the necessary validation. Bitcoin.org’s contemporaneous alert estimated that roughly half the network’s hash rate was mining this way. It documented a six-block invalid branch on July 4 and a three-block branch on July 5.
The incentive behind the shortcut was understandable. Mining is competitive. A miner learning about a new block wants to start working on its successor immediately. Waiting for data and validation costs time that competitors might already be spending on the new chain tip.
Those seconds do not give a miner cumulative progress toward solving a puzzle. Each hash is another attempt. But more time hashing on the branch that ultimately survives means more opportunities to earn a reward.
Some miners accepted the risk of starting before they had checked the foundation beneath them. In 2015, that risk became a loss.
There is an essential distinction here. Bitcoin does not physically prevent anyone from hashing on an unverified header. You can point mining hardware at whatever candidate you want. The question is whether the resulting block belongs to a chain that validating nodes will accept—and whether the reward will survive.
The new proposal tries to change the consequences of mining ahead of validation. That is a much bigger ambition than making a mining operation a little faster.
Consider what happens when a miner receives a new block under Bitcoin’s familiar model.
Its node needs enough information to reconstruct the block, check the relevant consensus rules, and determine the resulting set of spendable coins. That gives it a verified foundation for constructing the next candidate.
A block is therefore more than a container of transactions. It changes the state on which later transactions depend. If Alice receives an output in one block, Bob may receive a transaction spending that output in the next. A mistake in the first block can invalidate everything built on that mistake.
This dependency is a major reason that simply increasing capacity creates difficult trade-offs. Larger blocks can demand more transmission, processing, and storage. Differences between network connections and machines can become more consequential. A well-connected miner may learn the new state sooner than a poorly connected competitor.
But the starting point needs some nuance. Bitcoin already overlaps work.
Under ** BIP152**, compact block relay lets nodes reconstruct blocks using transactions they already possess, requesting missing transactions when necessary. Its high-bandwidth mode permits relay before all transaction validation has finished. Bitcoin Core’s explanation is explicit that nodes still complete validation before adding the block to their local blockchain.
The current system has already reduced how much data must travel at the moment a block is found. The research proposal goes further by changing how mining and transaction-state acceptance fit together across several blocks.
Think of a factory whose next production stage depends on an inspection. Improving the inspection helps. Starting another independent batch while that inspection runs can help too. The difficult word is independent. If every new product uses parts from the batch still being inspected, one failure can spread throughout the factory.
Bitcoin has the same problem in a much less forgiving form. The parts are spendable outputs. People may already be treating them as money.
