Ethereum: the power and price of programmability

Wonder

1. Why does the ability to execute arbitrary code inevitably expand the surface area for theft?
2. How did a community fork its own chain to reverse a hack, and why did Bitcoin reject that precedent?
3. What is the precise mechanism by which Bitcoin verifies heavy computation without executing it?

Explore

1. Ethereum launched in July 2015 with a goal distinct from Bitcoin: general-purpose programmability. While Bitcoin’s script was designed for specific, auditable financial transactions, Ethereum introduced a Turing-complete virtual machine (the EVM). This allowed developers to write and deploy any kind of program directly on the chain. The result was an explosion of tokens, decentralized finance protocols, and non-fungible tokens that were impossible to replicate on Bitcoin’s limited script environment.
2. The power of general computation came with a price. In June 2016, a recursive bug in The DAO smart contract allowed an attacker to siphon away roughly one-third of the funds raised. The Ethereum community faced a crisis: should they let the code execute as written (‘code is law’) or intervene? They chose intervention, executing a hard fork that moved the stolen funds to a refund contract by an irregular state change. This decision split the chain into Ethereum and Ethereum Classic, proving that social consensus could override immutable code.
3. The DAO incident highlighted a structural vulnerability: expressiveness enlarges the attack surface. Most subsequent thefts on programmable chains did not come from cryptographic breaks but from retained administrative keys, upgradeable contract logic, or simple coding errors. When a chain allows any program to run, every new feature adds complexity that can be exploited. The burden of security shifts from the protocol designers to thousands of individual developers who may lack rigorous auditing.
4. One answer to that attack surface came from inside the language itself. Move, written at Facebook for its abandoned Libra (later Diem) project, treats tokens as resources: a value the type system will not let a program copy or silently drop, so whole classes of double-counting and lost-funds bugs fail to compile rather than fail in production. When Diem was shut down in 2022, former Diem engineers carried Move to new chains. Aptos, founded by Avery Ching and Mo Shaikh, launched its mainnet on 17 October 2022 with Block-STM, a parallel execution engine that runs transactions optimistically, checks afterwards which ones touched the same data, and re-executes only those. The engineering was cutting edge; the launch was not. Exchanges listed the APT token before its distribution was published, the figures showed roughly half the supply going to investors, core contributors and the foundation, early on-chain activity fell far short of the advertised throughput, and the team conceded it “could have gone better.” Sui, built by another group of ex-Diem engineers, followed with its own Move-based mainnet in 2023. The lesson cuts both ways: safer languages narrow the attack surface, but technology does not rescue a launch that holders cannot see into.
5. Bitcoin’s response has been selective adoption rather than wholesale copying. Instead of general computation, Bitcoin has integrated specific, auditable capabilities. Taproot, activated in November 2021, introduced script trees that allow for complex spending conditions while maintaining privacy and efficiency. Wallets adopted PSBT (Partially Signed Bitcoin Transactions), a standard format for passing a transaction between signers. These additions enhance functionality without introducing a general-purpose virtual machine where any buggy code can compromise the entire network.
6. For heavy computation that cannot be handled by simple scripts, the BitVM approach offers a path forward. Proposed by Robin Linus in 2023, BitVM keeps heavy computation off-chain; only a disputed step is checked on Bitcoin, and only if someone challenges. This maintains Bitcoin’s lean base layer while enabling advanced applications, ensuring that the core protocol remains simple, verifiable, and resistant to the kind of cascading failures seen in general-purpose smart contract environments.
7. The trade-off is clear: Ethereum prioritizes flexibility at the cost of security complexity and upgradeability risks; Bitcoin prioritizes stability and explicitness by restricting computation. For any new feature added to a chain, the test is whether it introduces hidden state or admin powers that can be abused. If a capability cannot be verified through simple, transparent rules without relying on trusted third parties or complex social consensus forks, it carries inherent risk. This distinction defines the current architectural divergence between the two leading chains.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• V. Buterin, Ethereum: A Next-Generation Smart Contract and Decentralized Application Platform (2014)
• P. Wuille, J. Nick & A. Towns, BIP-341: Taproot (2020)
• R. Linus, 'BitVM: Compute Anything on Bitcoin' (2023)
• Contemporary reporting on The DAO hack (June 2016)
• Aptos Foundation, Aptos Tokenomics Overview (2022)
• R. Gelashvili et al., "Block-STM: Scaling Blockchain Execution by Turning Ordering Curse to a Performance Blessing" (2022)
• Contemporary reporting on the Aptos mainnet launch and tokenomics (CoinDesk/Yahoo Finance, October 2022)