The Blockchain Scalability Trilemma
Pick two: decentralization, security, or scalability — and design around the trade-off.
- Difficulty
- Moderate
- Time to result
- ~ongoing to results
- Steps
- 7
- Confidence
- 75%
Proposed by Vitalik Buterin, the scalability trilemma is a mental model stating that a blockchain architecture struggles to maximize decentralization, security, and scalability (transaction throughput) all at once. Most base-layer designs implicitly choose to protect decentralization and security while sacrificing raw speed, which is why global-scale computers like Ethereum process only tens of transactions per second versus tens of thousands for centralized payment rails. The practical use of the model is to deliberately push non-critical, high-volume activity onto faster, less-decentralized or less-secure secondary layers (side chains, shards, or separate execution-layer chains), while reserving the maximally secure and decentralized main chain for final settlement of higher-value activity.
Origin
Extracted from Acquired
Core principles
- 01You generally cannot maximize all three properties simultaneously on a single chain.
- 02Not every transaction needs the same level of decentralization/security, so tiering is rational.
- 03Execution and settlement can be separated into different layers, similar to real-time trading vs. end-of-day settlement in traditional finance.
How to run it
- 1
Name the three properties
Explicitly identify decentralization, security, and scalability as the three competing goals for the system you're designing.
- 2
Map your current position
Assess where your existing chain or application sits on each axis today (e.g., Ethereum at 15-45 TPS is highly decentralized/secure but low throughput).
- 3
Identify what the use case actually needs
Determine whether a given transaction type truly requires maximum security and decentralization or can tolerate more risk for more speed.
- 4
Push high-volume, low-stakes work off the main chain
Route large batches of routine transactions to side chains or shards that trade off some decentralization/security for throughput.
- 5
Separate execution from settlement
Let fast layers handle real-time execution, then batch and finalize results back to the most secure main chain, mirroring exchange trading vs. end-of-day settlement.
- 6
Reserve the base layer for final truth
Keep the main chain as the ultimate source of settlement truth for high-value transfers, since it retains maximum security and decentralization.
- 7
Re-evaluate as the network upgrades
Revisit the trade-off as protocol upgrades (like moving to proof of stake or adding sharding) shift the achievable frontier of the trilemma.
In the wild
Ethereum's planned sharding approach batches transactions off the main chain into ~64 shards and brings the results back as a single settled transaction.
→ This is designed to relieve congestion while still settling on the more secure main chain.
A project handles fast, low-stakes actions like likes/upvotes on Solana, then settles on Ethereum for higher security.
→ Demonstrates using two chains to hit different points on the trilemma for different transaction types.
Common mistakes
Expecting to max all three at once
Assuming a single-chain design can be maximally decentralized, secure, and fast simultaneously ignores the trilemma's core trade-off.
Treating all transactions equally
Routing trivial, low-value transactions through the same maximally secure/decentralized path as high-value settlement wastes capacity and drives up costs.
Underestimating implementation time
Solving even one side of the trilemma (e.g., proof of stake, sharding) has taken Ethereum years longer than originally planned.
Is it for you?
Best for
Protocol designers and technical strategists deciding how to architect a decentralized system under real throughput constraints.
Not ideal for
Centralized systems with a single trusted operator, where decentralization isn't a design goal at all.
From the transcript
“Yeah, let's introduce the scalability trilemma. So this was an interesting concept that was proposed by uh Vitalik, where he basically said, look, there are…”
“It's almost like not all laws are the Ten Commandments, like things can be treated with different weights.”
From the episode
Ethereum