The Verification Flywheel
Keep participation cheap enough to be a default habit, then defend the norm.
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 7
- Confidence
- 72%
Decentralization is not a parameter you set once; it is a habit that has to stay cheap. This model treats it as two reinforcing factors. The first is technical feasibility: can an ordinary participant verify the system on ordinary hardware, easily enough that they keep doing it without thinking about it. The second is culture: does the community treat independent verification as an ingrained responsibility worth defending. Each factor feeds the other, because a culture that values running nodes will fund the protocol work that makes nodes cheaper, and cheap nodes make that culture easy to sustain. The failure mode is subtle. When participation is merely possible rather than easy, there is constant pressure to save time and point at somebody else's node, and once the norm collapses it is very hard to bring back. So changes that buy throughput by raising participation cost get refused, even when they look free.
Origin
Extracted from Naval. Vitalik Buterin answers why Ethereum insists a node must run on a laptop, splitting the answer into technical feasibility and culture and describing how the two feed each other.
Core principles
- 01A norm survives only when the behaviour it demands is nearly free.
- 02Possible is not the same as default; possible degrades into delegation.
- 03Technical affordance and culture reinforce each other in both directions.
- 04A lost participation norm is far harder to rebuild than to protect.
- 05Subsidised participation is not evidence that the norm exists.
- 06Refuse performance gains that are purchased with participation cost.
How to run it
- 1
Name the participation act
Identify the specific thing ordinary members must do for the system to stay trustworthy. In a blockchain that act is running a node and verifying blocks locally rather than trusting somebody else's answer.
Pro tip State the act in one sentence a newcomer can follow. If it takes a paragraph, it is already too expensive.
- 2
Measure its true cost
Add up hardware, storage, bandwidth, setup time, and ongoing attention. The relevant question is not whether a determined expert can pay that cost, but whether a casual participant will keep paying it indefinitely.
Watch out Costs that fall on storage and maintenance are easy to overlook and are usually what pushes people into delegation.
- 3
Engineer the cost below the hobby threshold
Invest protocol work specifically in making participation cheaper, not only faster. Stateless clients are the archetype: they aim to let someone verify the chain while holding only a tiny amount of data locally instead of half a hard drive.
Pro tip Treat every reduction in participation cost as a security upgrade and budget it like one.
- 4
Make the cheap path the default
Verification must be what happens automatically, not what a conscientious user opts into. If the default is to trust another node, the population that actually verifies shrinks to whoever is unusually motivated.
Watch out When participation is hard but possible, even consensus participants drift toward pointing at the same shared node, which recreates a single point of failure.
- 5
Cultivate and name the norm
Say out loud that independent verification is a responsibility of membership, and celebrate the people who carry it. Culture is the half of the flywheel that cannot be shipped in a release.
Pro tip Point to communities that already hold the norm strongly as proof it is achievable rather than idealistic.
Watch out A norm that has decayed is far harder to restore than one that was never lost, so protect it before it is visibly failing.
- 6
Refuse cost-raising upgrades
When someone proposes multiplying the parameters or requiring far larger hardware, weigh it against the participation it prices out. Core contributors should be willing to actively resist such changes even when the performance gain is real.
Watch out These proposals always arrive framed as pure upside, because the cost lands on future participants who are not in the room.
- 7
Stress-test the norm without subsidies
If participation is currently sustained by grants or foundation programs, model what happens when they stop. A subsidised participation rate tells you about the budget, not about the culture.
Pro tip Track the ratio of unsubsidised to subsidised participants over time; it is the honest health metric.
In the wild
Ethereum holds a hard cultural line that the chain must be verifiable on a laptop, and core development actively resists proposals to multiply the parameters or require very large drives. The technical half of the flywheel then follows the cultural half: protocol work such as stateless clients is aimed at letting a participant verify the chain while storing only a tiny amount of data locally, where a node today can consume half a hard drive. Cheaper nodes make the norm easier to keep, and the norm keeps funding cheaper nodes.
→ Participation cost trends downward over successive upgrades instead of upward, so the verifying population can keep growing rather than consolidating.
Newer chains with heavier hardware requirements often maintain healthy-looking node counts through foundation subsidies and active campaigns encouraging people to run nodes. The flywheel test asks a different question: if the subsidies disappeared, would the network still work that way. Where participation is expensive and paid for, the observed count measures the size of the grant program rather than the strength of a norm, and the population can contract sharply the moment funding changes.
→ The apparent decentralization is revealed as a budget line rather than a habit, which changes how much assurance the chain can honestly claim.
Common mistakes
Confusing possible with default
Designers check whether a motivated expert can still participate and conclude the property is intact. In practice, if participation costs meaningful effort, ordinary members delegate to somebody else and the verifying set silently shrinks.
Buying performance with participation cost
Raising hardware requirements is the easiest way to increase throughput and the least visible way to centralize. The gain shows up immediately in benchmarks; the loss shows up years later as a smaller, more collusion-prone set of verifiers.
Reading subsidised participation as culture
Paying people to run nodes produces the same headline number as a genuine norm. The two only diverge when the funding stops, which is exactly when the system most needs the participation to hold.
Is it for you?
Best for
Protocol designers and community stewards responsible for keeping a decentralized system genuinely decentralized over years.
Not ideal for
Systems that are openly and deliberately centralized, where broad independent verification is not a design goal.
From the transcript
“It's once it gets to the point where running a node is so easy that people are comfortable doing it by default”
“Another important aspect is culture. People need to feel like it is this very ingrained responsibility that there's lots of this very independent verification going…”
“The two things play into each other. If you have a culture that values running nodes, then you have a culture that's going to value…”
From the episode
Vitalik: Ethereum, Part 1