What a contract cannot enforce
Three limits no code removes: data from outside, claims against people, and anything someone does off the chain.
This lesson has had no expert review. It was written for this platform and against the evidence it cites; nobody has gone through it independently.
Learning objectives
- You can name the three limits of enforceability and illustrate them with an example.
- You can identify where a chain of contracts becomes reliant on trust again.
Check your prior knowledge
Answer these for yourself before reading on. Wherever you hesitate is where this lesson pays off.
- How does a program on a chain know what a currency currently costs?
- What happens when someone breaks a promise made off the chain?
- Who does a claim run against when a program executes faultily?
Core concept
A contract sees nothing outside the chain
A smart contract can only react to data somebody has written to the chain. Prices, interest rates, weather, delivery status — none of it is there by itself. Whoever enters it makes a decision about what counts as true. Enforcement by code therefore ends exactly where the data comes from: the contract reliably enforces what it was told.
Code compels state changes, not human actions
A contract can move balances it holds itself. It cannot make anyone deliver goods, vacate a flat or make a payment off the chain. As soon as an arrangement requires an action in the world, it needs again what it was supposed to replace: someone who stands behind it, and a way to hold them to it.
Faultless execution is not the same as the intended outcome
A contract executes what is written in it — including when nobody wanted the outcome. A typo in a parameter, a condition that bites differently in a rare case, a price entered wrongly for a few seconds: in all of these the software worked correctly. There is then no authority that reverses the transaction because it was obviously unintended.
The limits are described, not merely asserted
That smart contracts have nameable limits — around external data, legal enforceability and the functions intermediaries perform — is described in the research literature and recorded here as evidence EVD-2026-0004. The record classifies mechanisms in general terms; it says nothing about any particular contract and is no substitute for a code audit.
Definitions
- Oracle in the glossary
- The path by which data from outside the chain reaches a contract.
- Enforceability
- The ability to bring about performance of an arrangement against the other side's will if need be.
- State change in the glossary
- A change to the data stored on the chain — the only thing a contract can bring about directly.
Model
Both sides inside the contract — enforced
One side needs a datum from outside — dependent on the source
One side requires an action in the world — dependent on a person
Outcome unwanted, execution correct — no authority reverses it
Worked example
The same arrangement, two endings
- Arrangement A
- swap token X for token Y, both on the same chain
- Arrangement B
- payout once a delivery has arrived
- Data needed for A
- none — both sides sit inside the contract
- Data needed for B
- somebody must enter “arrived”
A needs no trust: the contract holds both sides and executes them together or not at all. B depends on a person or service reporting the state — and therefore on their honesty and availability.
Only A is enforced by the code. B is executed by the code but determined by a report.
Reading: So the useful question is never “is it a smart contract” but: do both sides of the arrangement sit inside the contract? As soon as the answer is no, trust is in play again somewhere, and one should be able to name in whom.
Retrieval
Exercise on real data
Read the guiding questions and mark those answerable only from information outside the code.
Dimension 4: smart contracts →For EVD-2026-0004, read the scope section: what exactly does the source not establish?
Evidence register →Application
Someone describes an application to you as “fully trustless”. Which three questions test that?
Related case studies
Institutional reading
- Bank
- Which of your controls assume a transaction can be reversed?
- Advisory
- How do you explain the difference between “executed” and “enforced” to a client?
Key takeaways
- A contract reliably enforces what it was told — scrutiny belongs with the source.
- Code moves state, not people; every action in the world needs a claim again.
- Correct execution and the intended outcome are two different things.
Evidence
- EVD-2026-0004
Annual Review of Financial Economics — Smart Contracts and Decentralized Finance