Bitcoin’s global network hashrate may seem like an objective metric, but researcher and Cypherpunk Jameson Lopp reveals measuring it precisely is deceptively tricky. In a recent Proof of Work (POW) Summit talk in Prague, Lopp described his “hunt for the real Bitcoin hashrate” by evaluating the accuracy of various estimation algorithms.
As Lopp explained, most hashrate estimates derive from blockchain data like difficulty targets and block times. However, he noted the volatility in estimates over shorter timeframes. “If you’re only using the past 10 blocks, the hashrate can appear much higher or lower than it is,” said Lopp.
By aggregating hashrate data reported directly from mining pools, Lopp created a benchmark for testing blockchain-based estimates. He found the commonly used 1,000 block (~1 week) estimate had just a 3.8% average error rate. Lopp then tried blending multiple estimates, optimizing for accuracy. His best algorithm used 10 estimates from 100 to 1,000 blocks
We współpracy z: https://cryptoslate.com/cypherpunk-jameson-lopp-sees-need-for-machine-learning-to-improve-bitcoin-hashrate-estimates/
