Bitcoin’s hashrate dropped significantly after data centers in multiple Chinese provinces shut down last week. Explosions in at least one of the country’s coal mines forced the shutdown. A concurrent steep drop in bitcoin’s price helped to fuel widely misrepresentative narratives about  the short- and long-term effects on miners from the acute decline in hashrate.  

What happened?

  • A coal mine in China's northwest Xinjiang region flooded, which led to power outages.
  • “Security incidents” in Shanxi, Guizhou and Xinjiang provinces over the last few weeks led to state imposed restrictions on large energy consumers, like bitcoin miners.
  • Bitcoin’s hashrate dropped significantly as several mining facilities in Asia turned off their machines.

Exactly how large the hashrate drop was caused a bit of controversy on Twitter. Several mining data providers offer a range of estimates on Bitcoin’s hashrate between April 15 - 17.

  • BitInfoCharts saw a roughly 9% drop from 148 EH/s to 135 EH/s.
  • MiningPoolStats saw a more than 20% drop from 165 EH/s to 130 EH/s.
  • Glassnode showed a much larger drop of 46% from 198 EH/s to 107 EH/s.
  • Coin Metrics also reported a steep 46% decrease from 199 EH/s to 107 EH/s.

Exactly which mining farms were affected by the energy restrictions isn’t clear. But which mining pools have miners in the affected regions is suggested by pool hashrate data.

  • Poolin, Binance, Huobi and 1HASH all saw hashrate drops that corresponded directly to the network’s overall estimated hashrate decrease.
  • Other pools like F2Pool and ViaBTC didn’t experience similar decreases, possibly signalling that they have previously moved machines out of the region.

Understanding hashrate.

Predictably, the data showing the most severe drops in hashrate received the most attention on social media and by news media reports. But why these numbers vary and how they are derived is widely misunderstood.

Hashrate is only an estimate. The number is extrapolated from the speed at which new blocks are added to the Bitcoin blockchain. And based on how different data providers calculate their estimate, hashrate numbers that surface online will vary.

Significant variance is also seen in the immediate recovery of Bitcoin’s hashrate on April 18.

  • BitInfoCharts saw a 6% recovery to 111 EH/s.
  • MiningPoolStats saw 4% recovery to 135 EH/s.
  • Glassnode showed a 37% bounce to 147 EH/s.
  • Coin Metrics showed a 36% bounce to 145 EH/s.

So what?

Understanding how hashrate is calculated and the significant variance between difference calculations allows the “real” hashrate to be understood as a range of estimates instead of an absolute fact about the state of the Bitcoin network. Diagnosing the health of the network based exclusively on hashrate estimates, moreover, yields a very incomplete and often incorrect perspective.

So, after the hashrate drop, how are miners and the network faring?

Miners are more profitable.

  • Transaction fees: Since fewer miners are attempting to solve blocks, the speed at which blocks of new transactions are being added to the network has slowed. Thus, the amount of transactions sitting in mempools is piling up. As a result, fees for transactions are increasing, boosting the fee revenue portion of each block for miners.
  • Block subsidies: A drop in hashrate does not immediately cause a corresponding drop in mining difficulty, since it only adjusts once every 2016 blocks (roughly once every two weeks). An unchanged difficulty level means the same amount of hashing work is required per bitcoin mined. So, despite the price drop, which pushes down the dollar value of the new coins, miners with their machines still online are expending the same amount of hashing power to claim the same amount of bitcoins.

Also, far from a warning signal for the network’s health, the hashrate drop caused a notable drop in predicted difficulty for the next adjustment showing that everything is functioning as intended. Right before the hashrate drop, Bitcoin’s most recent adjustment saw difficulty increase by nearly 2%. For the next adjustment in less than 2 weeks, however, the predicted change will see difficulty drop by over 5%.

Was the drop in hashrate unexpected? Yes. Are reports of the decline sensationalized and somewhat misrepresentative? Yes. But will the network continue to function as it’s designed? Also yes.