Mining networks keep block production close to a target interval even as the amount of hardware pointed at them changes enormously. They do this by adjusting difficulty, and the choice of which variable to move is deliberate.
Two things could absorb a change in mining power
When more machines join, blocks are found faster. Something has to give, and the design could either let blocks arrive more quickly or make them harder to find.
Letting them arrive faster would accelerate issuance, so the supply schedule would depend on how much hardware the world happened to point at the network.
Adjusting difficulty keeps the interval steady, which keeps issuance on schedule. Predictable supply was the priority, so difficulty is the variable that moves.
The adjustment is arithmetic, not judgement
At fixed intervals, nodes compare how long the last stretch of blocks actually took against how long it should have taken.
The target is scaled by that ratio, within limits that prevent a single adjustment from swinging too far, and every node computes the same new value from the same chain data.
No vote or authority is involved. The adjustment is part of validation, so a block that used the wrong difficulty is simply rejected.
Miner economics follow the target
Difficulty determines how much work is required per block, which sets the electricity cost of producing one.
Rising prices attract hardware, which raises difficulty, which raises the cost of a block until it approaches the revenue. Falling prices push out the least efficient operators, which lowers difficulty and reduces the cost for those who remain.
The mechanism therefore acts as a feedback loop that pulls mining margins back towards the cost of the marginal operator, wherever that operator is and whatever they pay for power.
Adjustment lag creates real stress
Because adjustments happen at intervals rather than continuously, a sudden loss of mining power leaves difficulty too high for the remaining machines.
Blocks then arrive slowly, transactions queue and fees rise, and the situation persists until enough blocks have been found to trigger the next adjustment.
This is why abrupt regional shutdowns produce a period of congestion before the network settles, and why the adjustment interval is itself a design trade between responsiveness and stability.
Reward reductions are scheduled separately
The block reward does change, but on a fixed schedule known in advance rather than in response to conditions.
Each reduction halves miner revenue per block overnight while difficulty stays where it was, so the least efficient operators become unprofitable immediately.
Their withdrawal lowers difficulty at the next adjustment, and the network settles at a new equilibrium with fewer, more efficient participants and a greater share of revenue coming from transaction fees.