What Is Decentralization, and Why Does Bitcoin Need It?

beginner
AI-assisted

AI helped with research and writing. The editor chose topic and angle, decided what stays in and approved the article. Sources at the end.

How we work

Decentralization means that no single company, server, person or country can stop a system or change its rules on its own. Bitcoin needs it because money that one party can switch off is only as reliable as that party, and Bitcoin was built so that no such party exists.

That does not make Bitcoin immune to failure. It changes where a failure can happen and how far it can spread. To see why that matters, it helps to start with the opposite of decentralization: the single point of failure.

What a single point of failure is

A single point of failure is one part of a system whose breakdown stops the whole system. It can be a machine, a building, a cable, a piece of software or a supplier. Everything else can work perfectly, and the system still goes down, because everything depends on that one part.

Most systems have such points, and most of the time nobody notices them. They become visible on the day they fail. On November 6, 2026, an equipment fault at one shared computer center took 128 Japanese savings banks offline from 1:27 p.m. to 7:05 p.m. local time, according to the operators. Branches, ATMs, online banking and transfers stopped at the same moment, because all 128 banks ran on the same system. The banks themselves were fine. Their common center was not. Our analysis of single points of failure looks at that outage and at similar ones in detail.

The opposite of a single point of failure is not a system that never breaks. It is a system in which the failure of any one part does not stop the whole.

Four places where single points of failure hide

Single points of failure rarely look like one big switch. In practice they tend to sit in one of four places.

Where it hides What it looks like A real example
One location Many services depend on one building, one power feed or one region A fire at the North Hyde substation in west London on March 20, 2025 cut power to Heathrow Airport, which stayed closed for most of the next day
One update channel A single change reaches thousands of machines at the same time A faulty CrowdStrike update on July 19, 2024 crashed about 8.5 million Windows devices, according to Microsoft
One copy of the same logic The backup runs the same software on the same data and fails in the same way On August 28, 2023, the main and the backup flight-plan system of UK air traffic control both stopped on the same flight plan, within about 20 seconds of each other
One supplier A service that looks independent relies on a single provider underneath Canada's Interac debit network depended on one telecom provider, Rogers, and went down across the country when the Rogers network failed on July 8, 2022

The four often stack, and they explain why backups do not always help. A backup only protects against the failure it does not share. A second set of servers in the same building shares the location. After a fire at a data center in Pangyo, South Korea, in October 2022, the messaging company Kakao admitted that some of its systems had been duplicated only inside that one site. A backup that runs the same code on the same input shares the logic, which is exactly what happened to UK air traffic control in 2023. Two connections bought from the same provider share the supplier.

Centralized, decentralized and distributed

Three words are often mixed up, and the difference matters.

  • Centralized means one point of control. One server, one operator or one decision maker runs the system. That is simple and fast, and the whole system stands or falls with that point.
  • Distributed means the work or the data is spread over many machines. A distributed system can still be controlled from one place. A large cloud provider runs data centers in many countries, and one company still decides what happens in all of them.
  • Decentralized means control itself is split among many independent parties, and none of them can decide for the rest.

A distributed system removes the single machine. Only a decentralized system removes the single decision maker. Bitcoin is both: its ledger exists in thousands of copies, and no one controls them.

Switch off one point and see what keeps running

Both networks have 11 computers. Only the shape of their connections differs. Tap or click a node to switch it off, or use the button.

Centralized

Every connection runs through one hub.

Still connected: 11 of 11

Decentralized

Each node connects to several others, and no node is in charge.

Still connected: 11 of 11

  • Active node
  • Switched off
  • Connection

How Bitcoin spreads control across four layers

Bitcoin is not decentralized in one place but in several layers, and each layer removes a different single point.

The rules: every node checks for itself

A full node is a computer that stores Bitcoin's entire transaction history and checks every new block against the same set of rules. A block that breaks a rule is rejected by every node, no matter who produced it. There is no central server that holds the ledger. Thousands of copies do, and anyone can add another one. How that checking works step by step is explained in how Bitcoin transactions are verified.

The rules themselves change only when the people who run nodes choose to adopt new software. Nobody can push a rule change onto them, which is why Bitcoin changes only by consensus.

Block production: miners compete, and the network adjusts

Miners bundle transactions into blocks and compete to add the next one by spending computing power, a process called proof of work. Anyone with the hardware and the electricity can join. If a large group of miners stops, blocks arrive more slowly for a while, but the network does not wait for anyone to fix it.

Every 2,016 blocks, roughly every two weeks, each node compares how long those blocks actually took with the target of 14 days. It then raises or lowers the difficulty so that blocks return to about one every 10 minutes. A single adjustment is capped at a factor of four in either direction. The role of miners explains the mechanism in more depth.

The system has been tested for real. After China banned mining in 2021, a large part of the world's miners went offline within weeks. On July 3, 2021, at block 689,472, the difficulty fell by 27.94 percent, the largest downward adjustment in Bitcoin's history. Blocks kept coming the whole time.

The software: no update arrives on its own

Bitcoin has no central update server. Every node operator decides which software to run and when to upgrade. A faulty release therefore cannot reach every machine at once, the way an automatic update can.

The weak spot sits one level deeper. Most nodes run the same program, Bitcoin Core, and that is the "one copy of the same logic" risk from the table above. It has happened. On March 11, 2013, a large block was accepted by the new version 0.8 and rejected by the older 0.7, and the chain was split for about six hours, or 24 blocks. Developers and the operators of the largest mining pools resolved it by moving back to the older version. In 2018, developers fixed a reported bug in Bitcoin Core that could have allowed new coins to be created out of nothing. It was never exploited.

Both cases show the limits and the strengths of the design. A shared bug can hit the whole network at once. No one, however, can force a fix or a rule change onto the nodes. They have to choose it.

Custody: who holds the keys

The protocol can run without a single point of failure, and a user can still depend on one. Bitcoin held on an exchange is a claim against that exchange. If its servers, its cloud provider or its compliance team stop, the user waits.

During the Amazon Web Services outage on October 20, 2025, some Coinbase customers could not log in or trade for a total of 3 hours and 17 minutes, according to the exchange, which also said all funds were safe. The Bitcoin network itself was not involved. Block explorer data shows 86 blocks added between the start and the end of the outage. Bitcoin in self-custody depends only on the owner's keys and on any node that will pass a transaction on. Self-custody best practices covers how to hold keys safely.

How someone gets bitcoin in the first place is a choice as well. Exchanges are the most common route, but not the only one. Bitcoin can be bought from another person directly, through peer-to-peer marketplaces, or earned as payment, and a transaction can be sent through any node on the network. No single gate decides who gets in.

Why a large mining pool is not a single point of failure

Most blocks are found by a handful of large mining pools. That can look like a single point of failure. In practice it works differently, and the difference shows how Bitcoin's design handles concentration.

A mining pool is a coordinator, not an owner. Individual miners connect their machines to a pool so that they earn small, steady payouts instead of waiting months for the rare block they might find alone. The pool assembles the block and splits the reward. The machines, and the electricity they use, stay with the miners.

That makes a pool's share borrowed. If a pool started to censor transactions, cheated on payouts or simply grew too large, its miners could point their machines at another pool by changing a single setting, without asking anyone. A pool also cannot change Bitcoin's rules. A block that breaks them is rejected by every node, whoever found it.

This is not theory. In mid-June 2014, the pool GHash.io briefly passed half of Bitcoin's hash rate. Miners were urged to leave it, and on July 16, 2014, the pool publicly committed to staying below 40 percent of the network.

The same logic runs through the rest of the network. Mining gathers where electricity is cheap, nodes gather where hosting is convenient, and most nodes run the same software. Each of these is a choice that the people involved can reverse without anyone's permission, and none of them can change the rules on its own. That is the difference between concentration and control.

Speed, energy and slow change

Decentralization asks something in return. Each of these costs comes with an answer, and in some cases the cost is the point.

  • Speed. A new block arrives about every 10 minutes, and large payments usually wait for several confirmations. The base layer is built for final settlement, not for every coffee. Everyday payments settle in seconds on layers built on top, such as the Lightning Network.
  • Energy. Proof of work uses a lot of electricity, and that cost is what protects the history. Anyone who wanted to rewrite past blocks would have to redo the work. The energy mix has been shifting. In a 2025 Cambridge survey of mining firms that together ran about 48 percent of the hash rate, 52.4 percent of their electricity came from renewables and nuclear power, while natural gas was the largest single source at 38.2 percent.
  • Flexibility. Miners can switch off within minutes, which makes them an unusual kind of electricity customer. During Winter Storm Fern in January 2026, US miners powered down while the grid was under strain, MARA alone about 770 megawatts. Where wind or solar produce more than the grid can take, mining can buy power that would otherwise go to waste. A 2026 study of an Irish wind farm found that adding mining could lift the farm's revenue by 32 percent, although mining on surplus power alone did not pay, as our analysis of that study shows.
  • Slow change. Because nobody can impose a rule change, improvements take years of discussion and voluntary adoption. For money, that is largely a strength. The rules someone relies on today, including the limit of 21 million coins, cannot be rewritten by whoever happens to be in charge tomorrow.
  • Responsibility. The flip side of no one being able to freeze your bitcoin is that no one can restore it. A lost key stays lost, and a payment to the wrong address cannot be reversed.

Why this matters most for money

For most systems, a few hours of downtime are an inconvenience. For money, a single point of failure is also a single point of control. Whoever runs it can, in principle, stop payments, freeze accounts or change the rules, and the reasons do not have to be technical. Our essay on who holds the switch when a bank fails follows that thought.

Bitcoin was designed so that this point does not exist at the level of the protocol. Whether it exists for a particular person depends on how that person holds their bitcoin.

Key Facts

Bitcoin adjusts its mining difficulty every 2,016 blocks, roughly every two weeks, so block times return to about 10 minutes even after many miners switch off.

→ See the full table

After China's mining ban, Bitcoin's difficulty fell by 27.94 percent at block 689,472 on July 3, 2021, the largest downward adjustment in its history, and blocks kept coming throughout.

In mid-June 2014, the mining pool GHash.io briefly passed half of Bitcoin's hash rate, and on July 16, 2014, after miners were urged to leave, it committed to staying below 40 percent.

In a 2025 Cambridge survey of mining firms running about 48 percent of the hash rate, 52.4 percent of their electricity came from renewables and nuclear power.

During Winter Storm Fern in January 2026, US miners powered down while the grid was under strain, MARA alone about 770 megawatts.

Frequently Asked Questions

No single party controls it, and that is the part that counts. Every full node checks the rules and the ledger for itself, and nobody can change them for everyone else. Some activity clusters, such as mining in a few large pools, but those clusters rest on choices that miners and users can reverse at any time, and none of them can change the rules.

A government can ban mining or exchanges inside its own borders, but it cannot switch off nodes and miners in other countries. When China banned mining in 2021, blocks kept coming, and the network's difficulty adjusted down by 27.94 percent in a single step so that block times recovered.

Blocks arrive more slowly until other pools fill the gap or the next difficulty adjustment makes blocks easier to find. The miners behind a pool own their machines and can point them at another pool.

Yes, on the base layer it does. A new block arrives about every 10 minutes, a pace chosen so that thousands of independent nodes can receive and check each block before the next one arrives. Faster everyday payments run on layers built on top, such as the Lightning Network.

Distributed means the work or the data is spread over many machines. Decentralized means no single party controls those machines. A large cloud is distributed but centrally controlled, while Bitcoin is both distributed and decentralized.

Sources

  1. 1.Bitcoin Core — pow.cpp, difficulty adjustment every 2,016 blocks
  2. 2.Satoshi Nakamoto — Bitcoin: A Peer-to-Peer Electronic Cash System
  3. 3.National Association of Shinkin Banks, Shinkin Kyodo Center, Shinkin Central Bank — Joint notice on the system failure of November 6, 2026 (Japanese)
  4. 4.Jiji Press via Nippon.com — System failure hits 128 shinkin banks in western Japan
  5. 5.NESO — North Hyde Review, Final Report
  6. 6.TechCrunch — Microsoft says 8.5M Windows devices were affected by CrowdStrike outage
  7. 7.UK Civil Aviation Authority — CAP2993, Independent Review of NATS (En Route) Plc's Flight Planning System Failure on 28 August 2023
  8. 8.CRTC — Assessment of Rogers Networks for Resiliency and Reliability Following the 8 July 2022 Outage
  9. 9.Vixio — Rogers outage met with ministerial action, Interac adds backup supplier
  10. 10.Asia Economy — Kakao's report on the Pangyo data center fire
  11. 11.The Block — Bitcoin records its largest difficulty drop ever
  12. 12.Bitcoin Wiki — BIP 50, the March 2013 chain fork
  13. 13.Bitcoin Core — CVE-2018-17144 full disclosure
  14. 14.Amazon Web Services — Summary of the Amazon DynamoDB service disruption in the Northern Virginia (US-EAST-1) Region
  15. 15.Coinbase — Retrospective: AWS outage impact and Coinbase next steps
  16. 16.mempool.space — Blocks 919,913 to 919,998, mined during the AWS outage of October 20, 2025
  17. 17.Bitcoin.com News — Bitcoin difficulty logs 11.16 percent reduction, largest drop since China's 2021 mining crackdown
  18. 18.Cointelegraph — Bitcoin difficulty drops 11 percent during Winter Storm Fern
  19. 19.b10c — Bitcoin mining centralization in 2025
  20. 20.The Guardian — Bitcoin currency could have been destroyed by '51%' attack (June 16, 2014)
  21. 21.Finance Magnates — Another 51% scare: Ghash.io reaches 47% hashrate, users urged to leave (June 8, 2014)
  22. 22.Ars Technica — Bitcoin pool GHash.io commits to 40% hashrate limit after its 51% breach (July 16, 2014)
  23. 23.Cambridge Centre for Alternative Finance — Cambridge Digital Mining Industry Report (2025)
  24. 24.Marcel Sarnecki, Niall Burke — Bitcoin mining as supply-side flexibility in Irish wind energy integration, Energy Economics (2026)

Not financial advice. CanoeBit publishes educational content only. Nothing here is a recommendation to buy, sell, or hold any asset.