How does proof of work discourage cheating?

Proof of work makes producing an eligible block require repeated computational attempts. In Bitcoin, miners vary candidate block inputs and search for a header hash meeting the difficulty target. Finding a qualifying result is costly; checking it is comparatively inexpensive. Difficulty adjustment regulates the expected rate of block production as computational participation changes.¹

Among competing valid histories, Bitcoin follows the branch with the greatest accumulated proof of work. An attacker seeking to replace recent history must produce a competing valid branch with sufficient work to become preferred. The deeper the target transaction, the more subsequent work normally has to be overcome. The original Bitcoin paper frames this as a probabilistic competition between honest and adversarial computation.²

Miners earn protocol rewards and transaction fees, while paying operating and capital costs. These incentives encourage participation, but “cheating would be expensive” is not a complete security analysis. The relevant questions include access to computation, attack duration, available rewards from an attack and the effect on the attacker's remaining assets.

What a majority attacker can and cannot do

A sufficiently powerful attacker may reorganise recent history, exclude transactions or reverse its own earlier payment by replacing it with a conflicting spend. This is the classic double-spending concern.

Hash power alone does not reveal private keys, authorise spending someone else's coins or make an otherwise invalid issuance acceptable to honest validating nodes. These actions fail validity rules even if attached to substantial work. Conversely, an attacker below a majority can still have some probability of a successful limited reorganisation. “51%” is not a boundary below which every attack becomes impossible.

Distinguish three quantities:

• The ongoing revenue available to honest miners.

• The cost and feasibility of obtaining or redirecting attack capacity.

• The potential economic benefit from a particular attack.

They are related, but they are not interchangeable. A large annual reward budget does not mean an attacker must spend that annual amount before attempting a short disruption. Nor does a fall in token price mechanically translate into an immediate proportional loss of security: miner responses, equipment economics and difficulty adjustment matter.

A simple timing example illustrates another distinction. Suppose a hypothetical network targets a ten-minute average block interval. If participating hash power falls by 20% while difficulty remains unchanged, its expected interval becomes:

10 / 0.8 = 12.5 minutes.

That is an expectation across many blocks, not a schedule for the next block. An institution using a fixed number of confirmations may experience a longer funding delay until the protocol's adjustment mechanism responds.

Mining pools add a further layer. The operator coordinating block construction and the owners supplying computational capacity may be different parties. Pool concentration and ownership concentration should therefore be assessed separately, including how quickly participants could redirect capacity if an operator misbehaved.¹