Blockchain Trilemma Scorecard
No mechanism scores perfectly across all three dimensions. This is the core insight students should leave with.
| Mechanism | Decentralization | Security | Scalability | Best For |
|---|---|---|---|---|
| Proof of Work (PoW) | High | Very High | Low (~7 TPS Bitcoin) | Store of value, maximum trustlessness |
| Proof of Stake (PoS) | Medium-High | High | Medium (~15-30 TPS base, higher with L2) | Smart contract platforms, balance of security + efficiency |
| Delegated PoS (DPoS) | Low (21-101 delegates) | Medium | High (~1,000-4,000 TPS) | High-throughput applications, enterprise blockchains |
| BFT-based | Low (validator set) | High (instant finality) | High (~10,000+ TPS) | Permissioned networks, financial settlement, IoT |
Model Arguments Per Team
Team 1: Proof of Work (PoW)
Opening statement model (2 min):
PoW is the battle-tested gold standard of consensus. Bitcoin has operated continuously since 2009 — over 15 years — without a successful 51% attack on its main chain. The energy expenditure is not waste; it is the physical cost that makes attack economically irrational. An attacker needs to outpace the honest network's cumulative computational work, representing billions in hardware and electricity. No other mechanism creates this objective, physics-grounded security guarantee.
- Permissionless — anyone with hardware can participate; no identity or stake required
- Objective finality over time — Nakamoto consensus, longest chain rule
- Battle-tested: 15+ years, Bitcoin ~$1T market cap secured by PoW
- No "nothing-at-stake" problem — computational work is consumed, cannot be reused
- Credible neutrality — no central validator list, no slashing authority
- Energy consumption: Bitcoin ~150 TWh/year ≈ small country. Counter: ~75% renewable energy use (Cambridge data 2023)
- Mining centralization: Top 4 pools control ~60% of Bitcoin hashrate. Counter: pools are voluntary — individual miners can switch pools instantly
- Slow throughput (~7 TPS). Counter: Lightning Network handles payments at millions TPS; base layer is settlement layer
Key data points: Bitcoin: ~600 EH/s hashrate (2024) ~150 TWh/year energy 7 TPS base layer 0 successful 51% attacks on Bitcoin
Team 2: Proof of Stake (PoS)
Opening statement model (2 min):
PoS achieves the same cryptoeconomic security guarantee as PoW — attacking the network must be more expensive than any potential gain — without burning physical energy. Ethereum's Merge reduced its energy consumption by 99.95% while maintaining and arguably improving security through economic finality. Validators who misbehave have their stake slashed, creating stronger punishment than PoW where an unsuccessful attacker merely loses electricity costs.
- Energy efficiency: Ethereum PoS uses ~0.01 TWh/year vs. prior 78 TWh/year (99.95% reduction)
- Stronger punishment for attacks: slashing destroys stake permanently; PoW attacker just loses electricity
- Economic finality: with enough validators, reverting a finalized block requires burning 33%+ of all staked ETH
- Lower hardware barrier to entry: standard hardware suffices, reducing centralization pressure
- Composable with L2 scaling: Ethereum PoS + rollups achieves 100,000+ TPS
- Wealth concentration: large holders stake more and earn more. Counter: minimum 32 ETH entry, staking pools democratize access
- "Rich get richer": compounding rewards. Counter: reward rates ~4% APY are modest; inflation is low
- Younger than PoW — less proven at scale. Counter: Ethereum has $300B+ secured since Sept 2022 Merge with no incidents
Key data points: Ethereum: ~1M+ validators post-Merge ~0.01 TWh/year energy 32 ETH minimum stake Slashing: validator loses 1/32 stake minimum per offense
Team 3: Delegated Proof of Stake (DPoS)
Opening statement model (2 min):
DPoS is the rational engineer's answer to the blockchain trilemma: accept a degree of centralization to unlock the throughput necessary for real-world applications. EOS processes 4,000+ TPS. TRON handles 2,000 TPS. Cardano's Ouroboros-based delegation achieves high participation while maintaining meaningful decentralization. By electing a known, accountable set of block producers, DPoS achieves transaction finality in seconds rather than minutes — critical for payment systems, gaming, and social applications.
- High throughput: EOS 4,000 TPS, TRON 2,000 TPS — enables real applications
- Fast finality: block producers known in advance, rounds short (0.5s blocks on EOS)
- Democratic: token holders continuously vote; underperforming delegates are voted out
- Energy efficient: no mining, smaller validator set means less overhead
- Governance built-in: the delegation mechanism is also a governance mechanism
- Cartel risk: EOS top 21 BPs have coordinated votes, collusion documented. Counter: transparent voting, community can respond
- Small validator set = attack surface. Counter: 21-101 known validators are actively monitored, reputation at stake
- Low voter participation in practice. Counter: delegation is continuous and easy; liquid democracy
Key data points: EOS: 21 block producers, ~4,000 TPS TRON: 27 super representatives, ~2,000 TPS Cardano: ~3,000 stake pools (more decentralized variant) 0.5s block time (EOS)
Team 4: Byzantine Fault Tolerance (BFT)
Opening statement model (2 min):
BFT consensus was solving distributed trust problems decades before Bitcoin. In Practical Byzantine Fault Tolerance (pBFT) and its modern derivatives (Tendermint, HotStuff, PBFT), finality is instant and absolute — once a block is committed by 2/3 of validators, it cannot be reverted. No probabilistic finality waiting for "enough confirmations." For financial settlement, supply chain certification, or healthcare records, this property is not optional. Cosmos and Binance Smart Chain demonstrate that BFT-based systems can scale to hundreds of TPS without sacrificing security.
- Instant, deterministic finality: no block reorganizations, no "waiting for 6 confirmations"
- Tolerates up to 1/3 Byzantine (malicious/offline) validators exactly
- High throughput: Cosmos Hub ~10,000 TPS, Binance Chain ~100-300 TPS
- Well-suited for permissioned networks: Hyperledger Fabric, Quorum (enterprise Ethereum)
- Communication-efficient modern variants: HotStuff (used by Diem/Libra) reduces message complexity to O(n)
- Scalability of validator set: classic pBFT is O(n²) messages — works for small sets (<100). Counter: HotStuff, Tendermint solve this with leader-based rotation
- Requires known validator identities — not fully permissionless. Counter: for enterprise use cases, this is a feature, not a bug
- Liveness vs safety: BFT systems halt rather than fork under network partition. Counter: for financial systems, halting is preferable to inconsistent state
Key data points: Cosmos Hub: Tendermint BFT, ~150 validators Hyperledger Fabric: enterprise BFT Tolerates up to f < n/3 faulty nodes 1-6 second finality (Tendermint)
Use Case Scenario Verdicts
Below are 6 scenario cards with recommended verdicts. Accept well-reasoned alternative answers — the trilemma means context determines the best choice.
Scenario A: Global Censorship-Resistant Store of Value
"A government wants to ban transactions from political dissidents. Which consensus mechanism provides the strongest resistance?"
PoW with open mining is the hardest to censor — anyone with hardware can mine. The energy commitment and permissionless nature make it uniquely resistant. PoS and DPoS have smaller validator sets that can be more easily pressured or identified. BFT requires known validators — worst choice here.
Scenario B: High-Frequency Decentralized Exchange
"A DEX needs to process 10,000 trades per second with sub-second finality to compete with centralized exchanges."
Both achieve the throughput and finality requirements. DPoS (EOS-style) or BFT (Tendermint-style) are the only realistic options. PoW at 7 TPS and PoS at 15-30 TPS base layer both fail. Accept either DPoS or BFT with good reasoning about trade-offs accepted.
Scenario C: International Central Bank Settlement Network
"Central banks from 20 countries want a shared ledger for interbank settlement. They are known entities and regulation-compliant."
Known validators, instant finality, no mining waste, and strong consistency guarantees make BFT ideal for regulated institutional use. The "permissioned" nature is appropriate since participants are identified institutions. Hyperledger Fabric or Quorum are real-world examples. PoW's permissionless nature is irrelevant and wasteful here.
Scenario D: NFT Gaming Platform for 5 Million Users
"A gaming company wants to issue in-game NFTs and process millions of microtransactions cheaply and quickly."
DPoS handles throughput natively. PoS (Ethereum) with L2 rollups also works. Both Immutable X (PoS + StarkWare) and WAX (DPoS) have proven game NFT deployments. PoW is prohibitively slow and expensive for microtransactions. BFT is viable but less common in gaming.
Scenario E: Open-Source Smart Contract Platform
"A developer platform needs to support DeFi, DAOs, and arbitrary smart contracts while remaining decentralized."
Ethereum post-Merge is the canonical example — large decentralized validator set, smart contract support, active ecosystem. DPoS could work but centralization risk is higher. PoW historically powered Ethereum but proved too slow and energy-intensive as the ecosystem grew. PoS hits the right balance for this use case.
Scenario F: Supply Chain Transparency for a Multinational
"A Fortune 500 company wants immutable provenance records for its supply chain, with read access for public auditors."
BFT consortium (Hyperledger, Quorum) is the enterprise standard — controlled writes, public reads, instant finality, no energy waste. PoS (Ethereum) works for public verifiability if truly open provenance is needed. PoW is wasteful. DPoS is acceptable. Award points for any well-reasoned answer that handles the "public auditor" read requirement.
Common Student Errors & Misconceptions
- Claiming PoW is obsolete after Ethereum's Merge: Bitcoin still uses PoW and secures ~$1T in value. PoW and PoS serve different philosophies, not just different eras.
- Confusing DPoS with PoS: DPoS has a fixed, elected set of delegates. PoS validators are stochastically selected. The governance and centralization implications differ significantly.
- Stating BFT "can't be decentralized": Modern implementations like Tendermint support 150+ validators. While smaller than PoW, it is not inherently centralized — it depends on validator selection.
- Ignoring the "nothing-at-stake" problem for PoS: This is a real design challenge. Ethereum addressed it through slashing. Students defending PoS should acknowledge and explain the solution.
- Claiming any one mechanism "wins" universally: The trilemma exists precisely because no single mechanism is optimal for all use cases. Award extra credit for arguments that acknowledge context-dependence.
- Using outdated Ethereum data: Pre-Merge Ethereum was PoW with 78 TWh/year. Post-Merge (since Sept 2022) it is PoS at ~0.01 TWh/year. Ensure students use current data.
© Joerg Osterrieder 2025-2026. All rights reserved.