Bitcoin's Spam War: The 2.5% Referendum Behind Dashjr's Pool Exit

The split wasn't quiet, and it wasn't small. On Aug. 29, Luke Dashjr — a veteran Bitcoin Core developer and co-founder of the mining pool OCEAN — resigned as chairman, chief technology officer and director, and the company bought back all of his equity. The joint statement attributed the breakup to "differing visions for the future of Bitcoin mining." The vision outlived the job: tomorrow, Sept. 1, Dashjr is backing a hard fork that would swap Bitcoin's proof-of-work to BLAKE2b, a mining algorithm no existing Bitcoin machine can run, and call the chain everyone else uses "Spamcoin."

A pool founder leaving is insider news. A Bitcoin hard fork is a headline. The useful fact is that they are the same event, and the whole thing already lost.

The fight was BIP-110, an "anti-spam" soft fork championed by Dashjr — a rule to cap the picture-and-text blobs (Ordinals inscriptions) he calls spam, so blocks would stay money-only. Its sponsors were so short of support that they cut the activation threshold from 95% of hashrate to 55%. When activation arrived in August, support peaked at 2.53%. A crew called Roughnecks mined the fork's only two blocks on OCEAN's platform and then quit; the chain froze while Bitcoin ran more than 240 blocks ahead.

Miners vote in hash, and every block is a ballot. By that count the "clean Bitcoin" rewrite went down roughly 97.5 to 2.5 — call it a referendum, or call it everyone else simply ignoring it. Either way, the answer was definitive.

The part that gets buried under the drama is why miners refused a purer chain. On a ledger, "spam" has a different job title: bidder. Bitcoin sells scarce block space in a fee auction — users bid, miners take the highest offers — and data-heavy inscriptions became reliable high bidders. Refusing their fees is a pay cut, and miners read their own payroll. Dashjr's side wanted the abundant thing, cheap data, banned, so blocks would be scarce, clean, and cheap again. The majority that actually mines voted for the scarce thing that pays money: the block space itself.

That is the context for why a founder ended up out of his own company. OCEAN launched in late 2023 on a $6.2 million seed round led by Jack Dorsey — non-custodial payouts, a system called DATUM that lets individual miners build their own block templates, and a brand promise of "full block-template control to individual miners." Nearly three years later it mined about 2.5% to 3% of network blocks — roughly 25 exahash, a rounding error next to the pools that dominate.

In August the mission overrode the market. OCEAN's default configuration leaned toward the BIP-110 side, and on Aug. 8-9 a configuration error silently routed 18 hours of miners' hash to the stalled minority chain — miners who had actively selected the main network. Its displayed hashrate roughly halved within a day, and reported figures put the eventual collapse above 96%; the pool promised about 0.3 BTC of rebates. Its own VP of engineering, Jason Hughes, had published in July that BIP-110 had less than a 5% chance and that every major pool was ignoring it. The miner community called the episode a hashrate "hijack" and demanded removal of the leadership, and Blockstream's Adam Back — who had earlier called the whole campaign "idiocracy" — wanted the losses clawed out of Dashjr's salary.

Even the movement's own leaders read the result honestly. "The lesson from BIP-110," wrote Matteo Pellegrini, a prominent proponent and founder of Club Orange, "is that 15-20% of the nodes is not enough to change Bitcoin consensus." A founder who had staked his pool's credibility on that losing appeal was no longer an asset to the company that owned it. So the equity was bought back, and he was sent on his way.

The exit was a divorce, but the mission did not die with the job. Dashjr's next act is CONVOY, which he calls a "second run at decentralized mining"; it has disclosed no operating pool, no miner transfers, no fees, no infrastructure, and there is no evidence any material hash followed him. The Sept. 1 fork itself ships without replay protection — sign a transaction on one chain and it can be replayed onto the other. Broken hardware compatibility is the point: BLAKE2b wipes out every existing machine, which Dashjr frames as a strike at ASICBoost, the efficiency shortcut he believes hands large miners an advantage.

For a holder, that is the only near-term cost this affair is likely to produce: an operational pothole, not a re-pricing. Custodians and exchanges generally sort out split coins, and a retail holder doing nothing usually comes out fine. The people who should slow down are those signing raw on-chain transactions into the fork window.

The marginal buyer didn't flinch anyway. Bitcoin trades near $77,800, up 22% in twenty days, with about 60% market dominance and the fear-and-greed gauge at 62. The largest U.S. spot bitcoin ETF took in roughly $3 billion over the past month while the fork chatter peaked — accumulation, not panic.

So what is the investment lesson, cut to the bone? It isn't about who signs a pool's paychecks. This fight was the first political surface of Bitcoin's actual unresolved economics: who pays for network security when the new-coin subsidy stops doing it. The 2024 halving cut the per-block subsidy to 3.125 coins; the next, due around 2028, halves it again. Every cut makes fee-paying demand for block space — the very "spam" this campaign wanted to ban — more structurally important to the machine that safekeeps over a trillion dollars. Early analyses warned inscription fees were too volatile to solve the security budget on their own, and that caution stands. The referendum that just ran was the first clean market vote on the question, and it was decisive: keep the bidders, keep the fees, let an auction price the space.

And notice the mechanism buried in the mess. Bitcoin is accused of never clearing — coins that never die, chains that never shut down. Here, in weeks, the market froze a fork, punished a pool, and pushed its most prominent champion out of his own company. The failure was priced in public, not postponed. That is the system doing its job. The argument itself won't go away: the next fee drought meets the next subsidy cut in 2028, and someone will run this referendum again. The scoreboard to watch from now until then is what miners actually earn in fees — and this skirmish is why you now know exactly where to look.