How difficulty adjustments actually hit a small miner
Every two weeks or so, a small miner's payout changes even though nothing in their garage changed. The machine is the same, the electricity rate is the same, the pool is the same, but the network difficulty adjusted, and the same hashrate now earns a different amount of coin. Most explanations stop at "difficulty up, earnings down." This guide does the arithmetic properly: what the adjustment actually recalculates, how a run of ordinary adjustments compounds against a fixed small hashrate, and what decisions a one-or-two-machine operator should make when a new epoch lands. Nothing here is financial advice, and nothing here predicts prices, difficulty math works the same in any market.
What an adjustment actually recalculates
Bitcoin targets one block every ten minutes. Every 2,016 blocks, roughly two weeks, the protocol compares how long the last 2,016 blocks actually took against the expected 20,160 minutes (2,016 × 10 minutes). If blocks arrived faster than schedule, difficulty rises for the next period by roughly the same proportion. If they arrived slower, difficulty falls. The change is mechanical: every node runs the same calculation and gets the same answer. Nobody votes on it, and nobody can appeal it.
Two properties of this mechanism matter to a small miner. First, a single adjustment is capped at a factor of four in either direction, at most +300% or −75%, so difficulty cannot legitimately explode overnight; the dramatic moves are bounded by design. Second, the adjustment responds to total network hashrate, not to anything about your setup. When large operations plug in new fleets, blocks arrive faster, the retarget notices, and your share shrinks, even though you did nothing.
The share math: your hashrate is fixed, the denominator isn't
A miner's expected earnings are governed by a fraction:
your share = your hashrate ÷ total network hashrate
The network produces roughly 144 blocks per day, carrying a fixed total reward. Your slice of that reward is your share. Difficulty is the network's way of tracking the denominator: when difficulty doubles, the same hardware's daily coin output halves, because the machine now represents half the share it used to. Notice what difficulty does not touch: your electricity consumption. A machine draws the same watts at any difficulty. The cost side of your ledger is difficulty-proof; the revenue side is not. That asymmetry is the whole story of small-miner economics.
This is also why "difficulty" and "price" must be kept separate in your head. Price changes what each coin is worth. Difficulty changes how many coins your machine gets. A rising price can mask a shrinking share for months, payouts look fine in dollar terms while the coin amount quietly decays, and then a flat price reveals the decay all at once. Track your earnings in coin per day, not dollars per day, and the difficulty signal stays visible.
The compounding squeeze: six ordinary adjustments, worked
Individual adjustments are usually small, a few percent. Small miners shrug at +4% and go back to sleep. That is a mistake, because adjustments compound multiplicatively, and they compound against a cost base that never moves. The table below is a worked model, not a prediction: it assumes difficulty rises exactly 5% per adjustment (a plausible, unremarkable pace when network hashrate is growing) and indexes a fixed machine's coin output to 100 at the start. The math is just the share formula applied repeatedly, output after each epoch is the previous output divided by 1.05.
| Adjustment | Coin output index | Change vs start |
|---|---|---|
| Start | 100.0 | — |
| After 1 (+5%) | 95.2 | −4.8% |
| After 2 | 90.7 | −9.3% |
| After 3 | 86.4 | −13.6% |
| After 4 | 82.3 | −17.7% |
| After 5 | 78.4 | −21.6% |
| After 6 (~12 weeks) | 74.6 | −25.4% |
Read that last row carefully. Six unremarkable adjustments, none of them news, none of them dramatic, and the machine earns a quarter less coin than it did three months earlier, while consuming exactly the same electricity. If the machine's margin over its power cost was 20% at the start, it is now underwater, and no single event ever announced the crossing. At a gentler +3% pace the same six epochs cost about 16% of output (index 83.7), slower, same direction. This is the mechanism behind the familiar small-miner experience of a machine that "just stopped paying" over one season. Nothing broke. The denominator grew.
The total-cost view makes the squeeze concrete. Take your machine's daily electricity cost from the formula in the electricity-cost guide (power draw × 24 × your rate). That figure is identical in every row of the table. Divide it by the coin output index and you get your cost per coin earned, which rises by about 34% across the six epochs (100 ÷ 74.6). Difficulty doesn't raise your bill, it raises your unit cost, which is the same thing wearing a disguise.
When difficulty falls: the other direction is real
The adjustment is symmetric, and small miners benefit when it runs in reverse. Difficulty falls when enough miners shut down that blocks slow below schedule. The clearest modern example came in 2021, when mining restrictions in China forced a mass exodus of hashrate and several consecutive difficulty decreases followed. Every miner who stayed online saw their share, and their coin output per day, rise automatically, at unchanged cost. A single −5% adjustment lifts a fixed machine's output by about 5.3% (100 ÷ 0.95).
There is a pattern worth understanding here: downward adjustments happen because mining became unprofitable for the marginal operators, who switched off. The adjustment is the network's compensation to whoever remains. The miners best positioned to collect that compensation are the ones whose costs were low enough, or whose machines were efficient enough, to survive the squeeze that caused the exodus. Difficulty, in other words, systematically transfers earnings from high-cost miners to low-cost miners over time. A small miner cannot control the transfer, but should know which side of it their electricity rate puts them on.
What to do at each adjustment: a decision framework
An adjustment is information, not an emergency. The useful response is a quick, repeatable check, ten minutes with your own numbers, once per epoch:
Step 1: Recompute your margin in coin terms. Take your pool's reported coin-per-day for the new epoch and your fixed daily electricity cost. Margin = (coin revenue − power cost) ÷ coin revenue, valuing coin at a price you would actually sell at. Use the trend across the last three epochs, not one noisy day.
Step 2: Sort yourself into a band, and act on the band, not on the headline percentage.
- Comfortably positive margin (roughly 30%+). Do nothing. Ordinary adjustments are noise at this level; the compounding table says you have several epochs of headroom. Re-check next epoch.
- Thin margin (roughly 0 to 30%). This is where efficiency moves belong: clean the machine, fix airflow, and consider whether your firmware offers an efficiency mode that trades some hashrate for a better watts-per-terahash, at thin margins, lowering the cost side often beats chasing the revenue side. Also decide now, in writing, what margin makes you stop, so a future epoch doesn't decide for you by drift.
- Negative margin. The machine now converts electricity into coin at a loss versus buying the coin. The honest options are pausing (costs nothing, reversible, and positions you for a downward adjustment or a price move) or exiting (selling the hardware while it still has resale value, see the exit guide). What the framework forbids is the default most people choose: running a negative-margin machine for months because turning it off feels like admitting defeat.
Step 3: Never make a hardware decision on one adjustment. A single epoch can be distorted by a lucky or unlucky two weeks of block times. Purchases and sales should key off the multi-epoch trend in your cost per coin, the number from the compounding table, not off this fortnight's percentage.
The honest bottom line
Difficulty is not the weather; it is a ratchet with a release valve. It trends upward because efficient new hardware keeps joining the network, it releases downward when high-cost miners capitulate, and it compounds quietly in both directions. A small miner's defense is not prediction, it is measurement: know your cost per coin, watch it epoch by epoch, and have your stop/pause/continue thresholds decided before the squeeze arrives. The operators who get hurt are rarely the ones who misread a single adjustment. They are the ones who never recomputed.