Hashrate and security

Wonder

1. How is a number representing quintillions of calculations per second derived from just two observable variables?
2. What would it actually take to rewrite a day of bitcoin’s history?
3. What makes a hardware investment that yields zero utility outside of one specific consensus algorithm a stabilizing force for that network?

Explore

1. Hashrate is the aggregate computational capacity of all miners performing proof-of-work, measured in exahashes per second (EH/s), where one exahash equals 10^18 hashes. By 2024, this figure reached hundreds of EH/s. Nothing counts hash attempts directly; explorers and analysts estimate hashrate from the current difficulty and the rate at which blocks arrive. Because block discovery is a stochastic process governed by probability, short-term hashrate figures fluctuate significantly with variance in block timing.
2. Protection against rewriting history is economic rather than cryptographic. Rewriting recent history requires an attacker to sustain more hashing power than the rest of the network simultaneously. This necessitates acquiring specialized hardware and energy at a scale that is prohibitively expensive to assemble and highly conspicuous to attempt. One way to price an attack, argued by the economist Eric Budish in 2018, is by the mining revenue the attacker’s hardware could otherwise have earned; on that view, security costs roughly what miners are paid.
3. The difficulty adjustment mechanism ensures that as hashrate increases, the target hash value decreases proportionally every 2016 blocks (approximately two weeks). This keeps the average block interval near ten minutes. Consequently, individual miner probabilities of finding a block diminish as network power grows, but the total work an attacker must match rises. The energy expended by miners serves as a verification cost that makes double-spending economically irrational for rational actors who have sunk capital into the infrastructure.
4. Hardware specialization deepens this economic coupling. Bitcoin ASICs are designed exclusively to compute double SHA-256 hashes; they can mine only coins that use the same proof of work, and have little value otherwise. This ties miners’ financial fortunes directly to the network’s health and security. If the network were successfully attacked, that hardware would lose much of its value. This alignment of incentives discourages attacks from large-scale operators who have invested heavily in purpose-built capital equipment that depreciates rapidly but commands high entry costs.
5. Older blocks gain immutability as time passes. Forking the chain at a recent height requires outpacing the honest network only for a short duration, which is feasible with sufficient capital. However, rewinding history by hundreds of thousands of blocks would require sustaining majority hashrate for years, accumulating energy costs that exceed any potential gain from stolen funds. The depth of confirmation thus serves as a proxy for finality: each additional block makes reversal exponentially less likely for an attacker with a minority of the hashpower, and adds about ten minutes of full-network work for one with a majority.
6. To verify these dynamics, open mempool.space or a similar explorer to view the current network difficulty and estimated hashrate. Compare these figures with data from one year prior to observe growth trends and retargeting effects. Note that while individual machines contribute discrete amounts of power, the network figure represents a continuous estimate subject to statistical noise, reflecting the collective but variable nature of distributed mining operations across global jurisdictions.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• A. M. Antonopoulos & D. A. Harding, Mastering Bitcoin, 3rd ed. (2023), ch. 12
• Budish, 'The Economic Limits of Bitcoin and the Blockchain' (NBER Working Paper, 2018)
• mempool.space mining dashboard