Module 3: How Blockchains and Transactions Work
How Decentralised Is It, Really?
Chapter 13 of 13 · about 6 minutes
13.1 A word everyone uses
By now, Mint had heard the word "decentralised" hundreds of times. Every project used it. Every website claimed it.
"Is everything in crypto decentralised?" Mint asked.
"Almost nothing is fully decentralised," said Hash. "And almost nothing is fully centralised either. Decentralisation isn't a yes or no. It's a set of questions, about many different layers. A network can be very decentralised in one way and quite centralised in another."
This final chapter of Module 3 brings everything together. It gives you a practical way to judge decentralisation claims, using what you've learned about nodes, validators, block building, Layer 2s, bridges, and oracles.
13.2 Why decentralisation matters
Decentralisation isn't a goal for its own sake. It matters because of what it protects.
Censorship resistance. No single party can stop your transaction from being included.
Resilience. The network keeps running if some participants fail or go offline.
Security. Attacking or rewriting the network requires compromising many independent parties.
Credible neutrality. The rules apply equally to everyone and can't easily be changed to favour a few.
When decentralisation is weak, these protections weaken too, and you're back to trusting a smaller group, the very thing crypto set out to reduce.
13.3 The layers to examine
| Layer | Question to ask | Why it matters |
|---|---|---|
| Nodes | How many independent nodes verify the chain, and how hard is it to run one? | More verifiers means fewer people need to be trusted |
| Block producers | How concentrated are miners or validators? | Concentration makes collusion and censorship easier |
| Client software | Does most of the network run the same software? | A bug in a dominant client can affect the whole network |
| Infrastructure | Do most users and validators rely on a few hosting or RPC providers? | Outages or pressure on providers can affect many users |
| Block building | How many builders produce most blocks? | Concentrated builders can influence ordering and censor transactions |
| Governance | Who decides upgrades, and can a small group change the rules? | Control over rules is control over the network |
| Layer 2 operators | Who runs the sequencer, and who can upgrade contracts? | Sequencers and upgrade keys are trust points (Chapter 8) |
| Bridges and oracles | Who verifies cross-chain messages and price data? | Small signer groups are frequent attack targets (Chapters 10 and 12) |
13.4 Concentration in practice
Here are some real patterns worth knowing about.
Mining pools. Bitcoin miners usually join pools to earn steadier rewards. A small number of pools produce most Bitcoin blocks. Individual miners can switch pools, which limits the pools' power, but concentration is still watched closely.
Staking providers. On Ethereum, a significant share of staked ETH runs through a few large staking services and exchanges. The community has long debated whether any single provider holding too large a share is a risk.
Client diversity. Ethereum deliberately supports multiple independent software clients. If one client with a large share has a bug, the others can keep the network running. Networks where nearly everyone runs the same client are more fragile.
Cloud hosting. Many nodes and validators run on a few major cloud providers, creating a shared point of failure.
Block builders. As Chapter 11 described, a small number of builders produce most Ethereum blocks.
A common way to summarise concentration is the Nakamoto coefficient: the minimum number of independent entities that would need to collude to control or disrupt a network. A higher number is generally better. It's a simplification, but a useful one.
13.5 Censorship, a real example
In August 2022, the US Treasury sanctioned Tornado Cash, a privacy tool on Ethereum. Afterwards, some relays and builders chose to exclude transactions related to sanctioned addresses from the blocks they built. For a period, a large share of Ethereum blocks were produced through these filtering relays.
Sanctioned transactions were still eventually included, because other builders and validators didn't filter them. But it was a clear demonstration that concentration in block building can translate into censorship of particular transactions. The US Treasury lifted the Tornado Cash sanctions in March 2025, after a court ruling that it had overstepped, but the episode remains the clearest example of how block-building concentration can enable censorship. The Ethereum community has since worked on designs to make censorship harder, such as mechanisms that let validators force certain transactions into blocks.
"So decentralisation is what made sure those transactions still got in," said Node.
"Exactly," said Hash. "Not every builder had to agree. That's the point."
13.6 Decentralisation as a spectrum
It helps to think of each network or application as having a profile, rather than a single score.
| Mostly decentralised | Partly decentralised | Mostly centralised | |
|---|---|---|---|
| Example characteristics | Many independent validators, multiple clients, community-driven upgrades | Decentralised base layer, but a single sequencer or upgradable contracts | A company controls operations, upgrades, and funds |
| Trust required | Low | Moderate, at specific points | High |
| Example in this series | Bitcoin and Ethereum base layers | Many Layer 2s, many DeFi protocols | Centralised exchanges (Module 1, Chapter 7) |
Being partly centralised isn't automatically bad. Centralised exchanges are useful. Early-stage rollups with training wheels can fix bugs quickly. The key is to know where the trust points are, and to size your exposure accordingly.
13.7 Spotting "decentralisation theatre"
Some projects market themselves as decentralised while keeping meaningful control. Warning signs include an admin key that can upgrade contracts instantly, with no delay. Watch for a token "governance" process where the team and early investors hold most of the votes. Be wary of a small multisig controlling bridges, treasuries, or oracle feeds, or of validators or sequencers that are all run by the founding team. Another sign is a "decentralised" application with only one website front end and one RPC provider.
None of these automatically make a project bad. But they tell you that your trust is placed in people, not just code.
13.8 Module 3 wrap-up
Mint, Node, and Hash met at the end of the module.
"When we started, I thought 'pending' meant 'loading,'" said Mint. "Now I know about the transaction lifecycle, gas and base fees, Bitcoin's UTXOs, nonces and stuck transactions, confirmations and finality, block explorers, the scaling trilemma, rollups and other scaling designs, bridges, MEV, oracles, and what decentralisation really means."
"And what's the biggest lesson?" asked Hash.
Mint thought for a moment. "Every convenience comes from trusting something. The question is always what, and how much."
Node smiled. "That might be the most useful sentence in the whole series so far."
"You now understand how crypto works under the hood," said Hash. "Next, we start studying how prices move. That means charts."
13.9 Try it yourself
Pick one blockchain or Layer 2 you use. Using the table in section 13.3, answer as many of the questions as you can, using official documentation and independent dashboards. Where does it look strongly decentralised? Where are the trust points? Write a one-paragraph summary of what you're trusting when you use it.
Key takeaways from this chapter
- Decentralisation isn't a yes or no. It's a set of questions across many layers.
- It matters because it provides censorship resistance, resilience, security, and credible neutrality.
- Examine nodes, block producers, client software, infrastructure, block building, governance, Layer 2 operators, bridges, and oracles.
- Mining pools, staking providers, cloud hosting, and block builders show real concentration in practice.
- After the 2022 Tornado Cash sanctions, filtering by some builders showed how concentration can lead to censorship, and how decentralisation limits it.
- Partly centralised systems can be useful, as long as you know where the trust points are.
- Watch for decentralisation theatre: instant admin keys, concentrated governance, and small multisigs.
Coming up in Module 4: Technical Analysis for Crypto. Candlestick patterns, support and resistance, trends, volume, indicators, timeframes, and the quirks of applying technical analysis to a market that never closes.