all learn

Mining difficulty: the network's thermostat

Difficulty is the number that keeps block production on schedule no matter how much hashrate shows up. How the target works, how adjustments react to hardware booms and busts, and why difficulty — not hashrate claims — is the ground truth.

2026-10-16·updated 2026-10-16·4 min·pow.family research#beginner#mechanics
01

One number, one target

Recall the lottery: a block wins if its hash falls below a target value. Difficulty is simply the inverse of that target — the higher the difficulty, the smaller the target, and the more guesses a winning ticket takes. Difficulty doesn't make hashing harder per guess; it makes winning guesses rarer.

Expected attempts per block ≈ difficulty × (the constant for the hash's output space). Double the difficulty and you need double the guesses — so at a fixed hashrate, blocks come half as often; at double the hashrate, they come on schedule again.

02

The thermostat: difficulty adjustment

Chains want blocks on a clock — one per 10 minutes, one per minute, whatever the target interval is. So every chain runs a difficulty adjustment algorithm (DAA): it periodically compares how fast blocks actually arrived against the target and moves difficulty the other way. Blocks too fast → difficulty up. Blocks too slow → difficulty down. Hashrate floods in, the thermostat cools it; miners leave, it heats back up.

How *quickly* the thermostat reacts is a design choice with real consequences:

Adjustment styleBehaviorTrade-off
Slow periodic (Bitcoin: every 2,016 blocks ≈ 2 weeks)deliberately sluggishstability over responsiveness — if 30% of hashrate vanishes, blocks run slow for days until the next retarget
Per-block / continuous (modern chains: LWMA, ASERT-style)every block nudges difficulty toward the observed ratekeeps block time tight through hashrate swings; small coins need this when their hashrate is volatile
03

Worked example

A chain targets 60-second blocks. Over the last adjustment window, blocks actually arrived every 50 seconds — too fast, meaning hashrate outgrew difficulty. The DAA scales difficulty by 60/50 = ×1.2. Next window, the same hashrate now produces blocks every 60 seconds again. If a price crash then ejects a third of the miners, blocks stretch to 90 seconds until the next adjustment relaxes difficulty by ×0.67.

04

Why difficulty is the ground truth

Reported "network hashrate" is usually an *estimate* back-computed from difficulty and block times — marketing pages can inflate it. Difficulty, by contrast, is written into consensus and visible in every block header. When a small coin claims enormous hashrate, check its difficulty: trivial difficulty means the "hashrate" is borrowed or fleeting (the merged-mining lesson shows exactly how that happens).

05

See it live on pow.family

Project dashboards chart difficulty next to network hashrate — watch the thermostat track hashrate moves in the live data, and see each coin's current block-time adherence for yourself.

Sources

Related
More explainers
updated 2026-10-19ASIC vs GPU vs CPU: three ways to hashupdated 2026-10-19Bitcoin block rewards: subsidy, fees, and the coinbase transactionupdated 2026-10-16Emission schedules: why coins halve, decay, or never stopupdated 2026-10-16Merged mining (AuxPoW): one electric bill, two coinsupdated 2026-10-19The mining budget checklist: the true cost of a rigupdated 2026-10-19Electricity costs: the kWh math that dominates miningupdated 2026-10-19The mining glossary: every dashboard word, definedupdated 2026-10-16Pool mining vs solo mining: the variance tradeupdated 2026-10-16What is mining? The hash lottery and the block rewardupdated 2026-10-19Algorithm compatibility: hardware must match the puzzleupdated 2026-10-19Bitcoin difficulty: the 2,016-block thermostatupdated 2026-10-19Home installation: prep and the safety checklistupdated 2026-10-16How pools work: shares, the payroll ledgerupdated 2026-10-16Mining algorithms: SHA-256, Scrypt, RandomX — and why hardware differsupdated 2026-10-19Reading electricity tariffs: energy, demand, and fixed feesupdated 2026-10-16Tail emission and dilution: caps vs forever-moneyupdated 2026-10-19Airflow and temperature: the intake-exhaust loopupdated 2026-10-16Auditing supply claims: three formulasupdated 2026-10-19Halving planning: model subsidy cuts without the folkloreupdated 2026-10-16Effort and luck: reading a pool roundupdated 2026-10-19Reading miner specifications (and their units)updated 2026-10-19Stratum: how your miner talks to a poolupdated 2026-10-19Time-of-use rates: scheduling mining around the clockupdated 2026-10-19Miner vs node: who actually enforces the rulesupdated 2026-10-19Your first miner: the go/no-go checklistupdated 2026-10-19Demand charges: you're billed on your peak, not just your kWhupdated 2026-10-19Monitoring baselines and safe restartsupdated 2026-10-16PPS, PROP, PPLNS: how pools payupdated 2026-10-16Hashrate: what those TH/s numbers actually meanupdated 2026-10-19Confirmations: why 'confirmed' is a probability, not a switchupdated 2026-10-19Mine or buy? The honest comparisonupdated 2026-10-19Miner efficiency: the J/TH that decides who survivesupdated 2026-10-19Securing the operation: management access and payout custodyupdated 2026-10-16Stale shares and latency: why ping barely mattersupdated 2026-10-19Buying used miners: inspection and seller checksupdated 2026-10-19Payout address safety: the one typo that costs everythingupdated 2026-10-19Measure your miner's real power drawupdated 2026-10-19Reading profit estimates, sensitivity, and payback honestlyupdated 2026-10-19Reconciling pool payouts: expected vs actually paidupdated 2026-10-19Network difficulty vs share difficultyupdated 2026-10-19Rejected shares: the troubleshooting ladderupdated 2026-10-19Site electrical review: before you add loadupdated 2026-10-16Choosing a pool: the actual checklistupdated 2026-10-19Cooling and heat: every watt becomes heatupdated 2026-10-19Uptime, availability, and reconciling the energy model
Mining difficulty: the network's thermostat · pow.family