Bitcoin Mining Rabbit Hole

Bitcoin miners feed transaction data through a mathematical function called SHA-256 billions of times per second, changing one variable number with each attempt, until the output meets the network’s current target — there’s no shortcut, only brute force.

They’re running a number through a mathematical function over and over and over again, changing the number slightly each time, hoping the output looks a specific way.

That mathematical function is called SHA-256. You feed it any input — a word, a number, a block of transaction data — and it spits out a 64-character string of letters and numbers. The catch is that the output is completely unpredictable. Change one character of the input and the output looks nothing like the original. There’s no pattern to exploit, no way to reverse-engineer it.

Bitcoin’s network says: find an input that produces an output starting with a certain number of zeros. The only way to find it is to try. So miners take the current block of transactions, attach a variable number called a nonce, run it through SHA-256, check the output, change the nonce by one, run it again, check again. Billions of times per second.

That’s it. That’s the puzzle. It’s not clever — it’s brute force at industrial scale.

---

Bitcoin mining is intentional brute force — SHA-256 can’t be reverse-engineered, which means the only way to rewrite Bitcoin’s history is to physically redo all the computational work that built it.

None. That’s the whole point.

SHA-256 is specifically designed so there’s no way to work backwards from the answer. No shortcut, no trick, no smarter approach. The only path to a valid output is through sheer volume of attempts. A miner with twice the hardware makes twice as many guesses per second — that’s the only advantage available.

This sounds like a flaw. It’s actually the security model. Because there’s no shortcut, the only way to rewrite Bitcoin’s history would be to redo all the computational work that produced it — billions upon billions of guesses, for every block, going back as far as you want to change. The past is protected by the cost of the electricity that built it.

The “waste” of guessing is the point. The work is the proof.

Which raises a question that sounds almost too simple: if miners can guess billions of times per second, who decides how many guesses it should actually take to win?

---

Satoshi Nakamoto chose 10 minutes as a deliberate tradeoff — fast enough to be practical, slow enough that news of each new block can physically reach every miner on earth before the next one is found.

Satoshi picked 10 minutes, and it was a deliberate tradeoff rather than an obvious answer.

Shorter block times mean faster transactions — sounds better. But they create a problem: network propagation. When a miner finds a valid block, that news has to travel to every other miner on the planet before they stop wasting time on the old block. That takes seconds. With 10-minute blocks, propagation delay is a small fraction of the block time — manageable. With 1-minute blocks, propagation delay becomes a significant percentage of the block time, which means accidental forks (two miners finding blocks simultaneously before news of the first has spread) happen constantly.

More accidental forks mean less security and more wasted work.

10 minutes isn’t magic. It’s the point where propagation delay stops being a serious problem, while still being fast enough for practical use. A reasonable tradeoff for a global system that nobody controls.

---

Every 2016 blocks (roughly two weeks), every Bitcoin node independently recalculates mining difficulty using one formula: new difficulty = old difficulty × (actual time ÷ target time) — no vote, no committee, same result everywhere.

Every 2016 blocks — roughly two weeks — every Bitcoin node independently runs the same calculation.

It takes the timestamp of the most recent block, subtracts the timestamp of the block 2016 blocks ago, and compares the result to what 2016 blocks should take: 20,160 minutes (2016 blocks × 10 minutes). If the last 2016 blocks only took 15,000 minutes, blocks are coming too fast — difficulty needs to increase. The formula is simple: new difficulty = old difficulty × (actual time ÷ target time).

Two things worth knowing. First, every node runs this calculation independently and reaches the same result — no committee, no vote, no central server deciding. Second, the adjustment is capped at 4× in either direction. Without that cap, a sudden crash in mining power could trigger a death spiral — massive difficulty drop, blocks flood in too fast, next adjustment overcorrects, blocks slow to a crawl. The cap keeps the system stable even through dramatic changes.

When China banned mining in 2021 and roughly half of global mining power went offline overnight, the adjustment kicked in and the network recovered within weeks. No administrator required.

The timing mechanics explain how the network regulates itself. But they skip over something more basic — the transactions miners are actually racing to bundle in the first place. Where do those come from?

---

Miners pull transactions from the mempool — a public pool of unconfirmed transactions held by every node on the network — sorting by fee rate so higher-paying transactions get picked up first.

From the mempool — Bitcoin’s waiting room.

When you send a Bitcoin transaction, it doesn’t go directly to a miner. It broadcasts to the network and lands in the mempool, a pool of unconfirmed transactions that every node maintains. Think of it as a public queue. Miners dip into that queue, select transactions, and bundle them into a candidate block.

Here’s where it gets interesting: miners don’t take transactions in order. They sort by feerate — fee per unit of transaction size. High-fee transactions get picked first because miners are economically rational. Your transaction competes with everyone else’s for limited block space. When the network is busy and the mempool is full, fees go up because people bid higher to jump the queue. When it’s quiet, cheap transactions go through fine.

This is also why transactions sometimes get “stuck” — they entered the mempool with a low fee during a quiet period, and now the network is busy and nobody’s picking them up. They’re not lost. They’re just waiting.

---

Miners can technically exclude any transaction, but refusing a high-fee transaction simply hands that fee to a competitor willing to include it — making sustained censorship economically self-defeating.

Technically yes. Economically, it’s self-defeating.

A miner can choose to exclude any transaction for any reason. There’s no rule forcing inclusion. But a transaction sitting in the mempool is free money for whoever eventually picks it up. A miner who refuses to include a valid high-fee transaction is handing that fee to a competitor.

In practice, for miner censorship to actually work — to truly prevent someone from transacting — you’d need a majority of miners to all agree to exclude the same transactions, indefinitely, while sacrificing the fees those transactions carry. That’s a costly coordination problem with no economic incentive.

Bitcoin was also designed with this attack in mind. If a censored transaction keeps raising its fee, the pressure on miners to include it grows. The one miner who breaks ranks and includes it captures the reward. The incentive structure pushes against sustained censorship.

Not impossible. Just very expensive and unstable as a long-term strategy.

The economics push against bad intentions. But there’s a completely different edge case — one that has nothing to do with cheating, happens by accident roughly once every week or two, and reveals something remarkable about how the network resolves conflict.

---

When two miners find a valid block simultaneously, the network temporarily splits in two — each half building on a different block — until the next block is found anywhere and breaks the tie automatically.

For a few minutes, it actually splits in two — and that’s fine.

Imagine a miner in North America and a miner in Asia both find a valid block at almost the same moment. The North American miner broadcasts to nearby nodes. The Asian miner does the same. Because of propagation delay — the physical time it takes data to travel through underwater cables — the two halves of the network hear about different blocks first.

For a short window, the network is genuinely divided. Nodes in the Americas think Block A is the latest. Nodes in Asia think Block B is. Both groups start building their next block on top of what they believe is the current tip.

Nothing is broken. No one panics. The network has a tiebreaker mechanism, and it kicks in automatically the moment anyone, anywhere, finds the next block.

---

Bitcoin resolves competing chains using nChainWork — a running total of cumulative computational effort — and the moment one chain has more accumulated work than the other, every node on the lighter chain switches instantly, with no timer and no human decision required.

There’s no timer. The logic isn’t based on minutes — it’s based on work.

Inside Bitcoin’s code is a variable called nChainWork. It tracks the total cumulative computational effort that went into building a chain from the very first block. Not the number of blocks — the actual hashing energy required to produce them.

When two competing chains exist, every node is constantly comparing their nChainWork values. The moment one chain has more accumulated work than the other, nodes on the lighter chain switch. Instantly. Automatically. No vote, no human decision.

In practice this usually resolves with the very next block — whoever finds it first tips the balance and the losing chain is abandoned within seconds of that block propagating. But it can technically last longer. If both chains keep finding new blocks at roughly the same time — pure bad luck — the race continues until one side pulls ahead. It gets less and less likely with each additional block that they stay tied, because now two consecutive simultaneous finds have to happen. The probability drops astronomically with each round.

The resolution is always the same: one chain accumulates more work. The other is dropped. The software never needed a timer because mathematics does the job faster and more reliably than any clock could.

---

Yes — a miner can find a valid block, watch it propagate across half the network, and lose the entire block reward minutes later when a competing chain becomes longer and their block is abandoned in what’s called a chain reorganisation, or reorg.

Completely. And it’s brutal.

You find a valid block. Your hardware lights up. Your pool dashboard shows the win. Your block propagates across half the network. For several minutes, half the world’s miners are building on top of your block.

Then someone on the other side finds the next block, and their chain is now longer. Every node that receives this new chain performs what’s called a reorg — a chain reorganisation. They disconnect your block, roll your transactions back into the mempool, attach the new chain, and carry on. Your block becomes a stale block. Your coinbase transaction — the 3.125 BTC reward you’d just “won” — vanishes from history.

You did valid work. You found a legitimate hash. You followed all the rules. It counts for nothing.

This is also why Bitcoin won’t let miners spend their block reward until 100 blocks have been built on top of it — a deliberate buffer to ensure the block is genuinely buried in history before the reward is treated as real.

---

Six confirmations became the standard because Satoshi’s original whitepaper showed that even an attacker controlling 30% of global hash rate has less than a 2% chance of reversing a transaction buried six blocks deep.

Satoshi worked out the probability that a slower attacker could catch up to a longer honest chain, and 6 confirmations is where that probability becomes negligible for normal transactions.

Accidental forks — two miners finding blocks simultaneously by chance — almost never go deeper than one or two blocks. For that to persist to six blocks, you’d need six consecutive ties, each resolved the wrong way. The probability is so small it’s essentially theoretical for a non-targeted attack.

Satoshi published the math in the original whitepaper. With an attacker controlling even 10% of hash rate, the probability of successfully reversing a transaction with 6 confirmations is less than 0.1%. With 30% of hash rate, it’s still under 2%.

Six became the standard. For large transactions — an exchange receiving a major deposit, a merchant selling something expensive — more confirmations add more certainty. For small everyday payments, even one or two confirmations is practically fine because the cost of a successful attack vastly exceeds the value being stolen.

---

Yes — in March 2013 a software bug split the Bitcoin network for 24 blocks (about six hours), requiring rare human coordination between developers and miners to resolve.

Once, in March 2013.

A software bug in different versions of Bitcoin caused a fork that lasted 24 blocks — about six hours. Half the network was on one chain, half on another. It wasn’t an attack, just a compatibility issue between an old version of Bitcoin’s software and a newer one that handled a specific database edge case differently.

The fix required something Bitcoin almost never needs: human coordination. Core developers got on IRC, diagnosed the problem, and asked miners to deliberately downgrade to the older version to consolidate the network. It worked, but it was uncomfortable — a rare glimpse of the social layer underneath the code.

Outside of that, deep multi-block reorgs on Bitcoin have been essentially nonexistent since the early days. The economic cost of maintaining a parallel chain long enough to cause a reorg of any depth is enormous, and it gets exponentially more expensive with every block.

That cost is enormous partly because of what mining became as an industry. It didn’t start this way — and the story of how it evolved explains why your gaming PC is now completely irrelevant to the whole operation.

---

Gaming PCs can no longer compete because Bitcoin mining is dominated by ASICs (Application-Specific Integrated Circuits) — chips built exclusively to run SHA-256, performing the equivalent work of thousands of GPUs at a fraction of the electricity cost.

Because Bitcoin mining hardware evolved to do exactly one thing, and nothing else.

In the early days, regular CPUs mined Bitcoin fine. Then people discovered GPUs — graphics cards — were much faster because they could run many calculations in parallel. GPU mining dominated for a while. Then FPGAs (programmable chips) edged them out. Then ASICs arrived and ended the competition entirely.

ASIC stands for Application-Specific Integrated Circuit. It’s a chip designed and manufactured to do exactly one computation — Bitcoin’s SHA-256 hash function — with no ability to do anything else. No browsing, no gaming, no video editing. Just SHA-256, billions of times per second, with maximum energy efficiency.

A modern ASIC does the same hashing work as thousands of GPUs, using a fraction of the electricity. Competing with ASICs using a gaming PC is like entering a Formula 1 race in a family sedan. You’re technically both cars. That’s where the similarity ends.

The downside of this specialisation is that ASICs are useless if Bitcoin fails — they have no alternative function. Miners are making a long-term bet with physical hardware, not just electricity.

Which creates a new problem: if competing means owning industrial hardware, how does anyone with just a few machines survive? The odds of winning a block solo become almost impossibly small.

---

When a mining pool finds a block, the reward splits proportionally by contributed hash rate — tracked via “shares,” which are partial solutions that prove each miner is working even when they haven’t found the winning hash.

Yes, but the split is proportional to how much work each miner contributed — not equal shares.

Solo mining is effectively a lottery. You might have 0.001% of global hash rate, which means you’d expect to find a block roughly once every few decades. That’s not a business. Pools solve this by having everyone work together: each miner points their hardware at the pool, the pool coordinates what everyone mines, and when any member finds a valid block, the reward gets distributed according to each miner’s contributed hash rate.

The pool tracks this using “shares” — partial solutions that prove a miner is working, even though they’re not valid enough to submit to Bitcoin’s network. A miner submitting many shares is doing a lot of work; a miner submitting few shares isn’t. Rewards flow proportionally.

Pools charge a small percentage fee — typically 1-3% — for running the coordination infrastructure. In exchange, miners get predictable, steady income instead of a lottery ticket.

The tradeoff worth knowing: pools are a mild centralisation pressure. Large pools control significant hash rate and could theoretically coordinate on what transactions to include or exclude. This is why the Bitcoin community watches pool concentration carefully, and why protocols like Stratum V2 aim to give individual miners more control over their own block templates.

Once you understand the scale of all this — thousands of specialised machines, running continuously, worldwide — one criticism becomes very hard to ignore.

---

Bitcoin mining energy isn’t wasted — it produces unforgeable proof of work, the mechanism that makes Bitcoin’s transaction history trustworthy without any central authority guaranteeing it.

The electricity bought something real — it just isn’t a physical object.

What Bitcoin mining produces is unforgeable proof of work. The only way to produce a valid block hash is to have burned the electricity. You cannot fake it, borrow it, or simulate it. The energy expenditure is permanently encoded in the output. This is why the chain’s history is trustworthy: rewinding it would require redoing all the work, at today’s energy prices, which gets more expensive the further back you want to go.

Compare this to a central bank. A central bank produces trust in a currency through legal authority — laws, armies, institutional reputation. Bitcoin produces trust through physics. Neither approach is “free.” Both cost something. The question is which cost you find more acceptable.

On the energy source question: miners are extremely price-sensitive because electricity is their main input cost. This pushes them toward the cheapest available electricity — which is often curtailed renewable energy that would otherwise be wasted. Hydroelectric dams in seasons of excess rainfall. Flared natural gas from oil fields that would otherwise burn into the atmosphere anyway. The map of mining activity roughly traces the map of stranded or surplus energy.

Is it still a lot of energy? Yes. Is it entirely waste? The security it purchases suggests otherwise.

---

When the last Bitcoin is mined around 2140, the block subsidy drops to zero and miners will depend entirely on transaction fees — a model that holds only if Bitcoin’s transaction volume has grown large enough by then to make fees economically sufficient.

This is the most serious long-term question in Bitcoin, and the honest answer is: we don’t fully know yet, but the design assumes transaction fees will fill the gap.

Right now miners earn two things per block: the block subsidy (newly created Bitcoin — currently 3.125 BTC, halving roughly every four years) and transaction fees (what users pay to have their transactions included). The block subsidy will reach zero around the year 2140. After that, fees are the only miner income.

The argument for why this works: as Bitcoin becomes more widely used, more transactions compete for block space, which drives fees higher. A future Bitcoin network processing enormous global transaction volume could generate enough in fees to make mining economically sustainable indefinitely, with no new supply required.

The argument against certainty: transaction fee revenue today is small relative to the block subsidy. We’re a long way from proving that fees alone can sustain miner security at the scale needed. The Lightning Network, which handles many transactions off-chain, also reduces pressure on block space — which could reduce fee revenue rather than increase it.

Satoshi was aware of this. The design assumes the problem will be solved by the time it matters, because Bitcoin’s adoption and usage will have grown to the point where fees are sufficient. Whether that assumption holds is one of Bitcoin’s open questions — and won’t be answered for over a century.

What we do know: each halving is a live experiment in how miners respond to reduced subsidies. So far, every halving has been absorbed. The industry adjusts. The network continues. The next full test isn’t for 116 years.

---

Related Deep Dive Threads

• Bitcoin Network →
• Bitcoin Blockchain →
• Bitcoin Regulation & Future →

Still curious? The transactions miners are competing to include have their own rabbit hole — what’s actually inside a transaction, how your wallet knows your balance, and why Bitcoin doesn’t work the way most people assume. That thread is here.