The Moment a Miner Throws Away Its Own Work
You find a valid block. Real electricity burned, nonce checked out, hash below target, timestamp clean. By every rule of the protocol, you won. Then you broadcast it, watch a competing block take the chain tip, and quietly abandon yours.
Not because you were cheated. Because you chose to.
This is voluntary self-orphaning, and it happens more than most guides admit. Understanding why requires peeling back what "winning" actually means in proof-of-work mining. The answer turns out to be less about individual blocks and more about the relentless arithmetic of expected value across thousands of them.
The Block Race Nobody Talks About
When two miners find valid blocks at nearly the same height simultaneously, the network splits briefly. Different nodes hold different chain tips. The tie-breaker, per Nakamoto consensus, is whichever chain accumulates more total proof-of-work first. Usually the next block settles it: one chain grows, the other dies, and the miner on the losing side collects nothing. The subsidy, the fees, all of it: gone.
Ordinary orphaning. Involuntary and painful.
Voluntary self-orphaning is a different animal.
Here's the scenario. A mining pool finds block N. Simultaneously, or within a few seconds, a competing pool broadcasts their own valid block N. Both legitimate. Race live. The first pool now has a choice: keep mining on top of their own block N, or switch to the competing block N.
Switching feels like surrender. It isn't.
If the competing block already has 40% of network hashrate building on it and the first pool controls only 15%, the probability math gets cold very quickly. The pool's block needs them to find block N+1 before anyone else does. The competing block has the rest of the network working for it. Sticking with their own block is, in expected-value terms, often the losing move.
The Arithmetic That Forces the Decision
Let's make this concrete. Total network hashrate: 600 exahashes per second. Pool A controls 60 EH/s, roughly 10%. They find block 840,000. While propagating, Pool B (also 60 EH/s) broadcasts a competing block 840,000. The rest of the network, 480 EH/s, starts receiving both.
Assume propagation is messy: 300 EH/s of the neutral hashrate latches onto Pool B's block before Pool A's block reaches them. Pool A is now racing with 60 EH/s against 360 EH/s building on Pool B's version. Pool A's odds: roughly 60/420, about 14%.
If Pool A abandons their block and joins the majority chain, they surrender the block reward entirely. But they immediately contribute to extending the dominant chain, and their future blocks are no longer at risk of being orphaned because they built on a likely-loser. They preserve their position where it counts.
The decision boundary is not sentiment. It is: expected revenue from fighting versus expected revenue from folding. When your block has already lost majority support, folding is rational.
When It Gets Strategic (and Uncomfortable)
That's the honest, defensive version. There's a more aggressive version the research community has documented, called selfish mining, first formally described by Eyal and Sirer.
The mechanic: a pool with sufficient hashrate finds a block and deliberately withholds it rather than broadcasting immediately. They keep mining privately, building a secret lead. When the public network catches up to their hidden chain, they release it, orphaning honest miners' work and collecting disproportionate rewards. It is, essentially, a chess clock used as a weapon.
This is voluntary orphaning of other people's blocks, using a withheld chain as the instrument. The pool sacrifices immediate propagation advantage to manufacture orphans for competitors. The original paper put the profitability threshold at around 33% of network hashrate, though later refinements pushed it lower under certain propagation assumptions.
Confirmed large-scale selfish mining on Bitcoin's mainnet has been hard to prove definitively. The incentive is real. The execution risk and reputational cost are also real. The part most casual explanations skip: it is not a bug anyone will patch. It is a property of the incentive structure, sitting there permanently.
What People Get Wrong About Orphan Rates
The common assumption is that orphaned blocks are failures, accidents, signs of a sick network. Low orphan rates are healthy, sure. But a miner voluntarily self-orphaning is actually the protocol working correctly. Rational actors updating on new information and switching to the dominant chain is exactly what keeps Nakamoto consensus stable.
The wrinkle people miss: orphan rates are not uniform across participants. A pool with fast, well-peered nodes in Frankfurt, Tokyo, and São Paulo will propagate blocks in under 100 milliseconds to most of the network. A pool running through a single data center with poor peering might take 800 milliseconds. In a race often decided in under two seconds, that gap is enormous. The slower pool faces more involuntary orphans and is more likely to find itself on the wrong side of a voluntary-orphan calculation.
This is why large pools invest heavily in relay networks like FIBRE (Fast Internet Bitcoin Relay Engine), which compresses and pre-announces block headers before the full block propagates. Infrastructure spending to win millisecond races that collectively determine millions in annual revenue.
The Deeper Lesson About "Winning" a Block
Finding a valid hash is necessary but not sufficient. The block has to survive long enough to be buried under subsequent work. A block found in isolation, propagated slowly, or found at the same moment as a faster competitor might be valid by every cryptographic measure and still earn nothing.
Consider two miners: Reza and Sofia both find block 840,000 within 400 milliseconds of each other. Reza's pool has superior relay infrastructure and 18% of network hashrate behind it. Sofia's pool has 8% and slower propagation. Within three seconds, 55% of the network is building on Reza's block. Sofia's pool switches. They orphan their own valid block, eat the loss, and move on.
Sofia's pool made the right call. Fighting would have been expensive pride.
Here's the question worth sitting with: does the protocol reward the best miner or the best-connected miner? Both, honestly, and the gap between them is smaller than it used to be but has not closed. The cryptographic lottery is fair. Everything around it, the infrastructure, the peering, the relay networks, the latency, is very much not. Voluntary self-orphaning is just the moment that reality becomes a line item.