Selfish Mining

Selfish mining takes advantage of a subtle flaw in how a proof-of-work network resolves competing chains: nodes accept whichever valid chain is longest, without asking how or when its blocks were found. An attacker who withholds a block instead of broadcasting it immediately can quietly extend a private lead, then reveal it at the moment it does the most damage to honest competitors.

The strategy was formalized in 2013 by researchers Ittay Eyal and Emin Gun Sirer in the paper "Majority is not Enough," which showed the Bitcoin protocol was not as incentive-compatible as widely assumed. Their model divides the network into an honest majority and a selfish pool controlling a fraction of total hash rate. Depending on how well the attacker's blocks propagate relative to honest ones, a pool with roughly 25 to 33 percent of network power can out-earn honest mining, well below the 51 percent usually assumed necessary to manipulate a chain.

Beyond direct profit, the paper's authors warned that selfish mining creates a centralizing feedback loop: rational miners are drawn to join the more profitable selfish pool, pushing it closer to majority control. The concern became concrete in 2014, when the Bitcoin pool GHash.io briefly exceeded 40 to 50 percent of network hash power, prompting some miners to voluntarily leave over decentralization fears.

Selfish mining remains largely theoretical in practice, since sustained execution risks reputational damage and requires precise timing, but it underscores why hash power concentration in a handful of pools is treated as a structural risk to proof-of-work security.