When nothing breaks

The class of incidents in which every contract did exactly what it says — and the outcome is a loss anyway.

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 distinguish economic exploits from logic errors and name what follows for the examination.
You can name the three ingredients such an incident needs.
You can explain why an audit report covers this class only partly.

Check your prior knowledge

Answer these for yourself before reading on. Wherever you hesitate is where this lesson pays off.

What does a code audit examine — and against what does it measure “correct”?
Where does a protocol get the price it computes with?
What does it cost to move a great deal of capital for the duration of one transaction?

Core concept

Fact

Correct and loss-making all the same

A logic error is a divergence between what the code does and what it should do. An economic exploit is something else: the code does what it should, but the assumptions under which that rule made sense do not hold at that moment. A logic error is checkable against a specification. An economic exploit is only checkable against a model of the circumstances — and that model is rarely written down anywhere.

Fact

Three ingredients

Almost every incident of this class needs the same things: a rule depending on an observed quantity (usually a price); a way to move that quantity briefly; and enough capital to do so. The third point is the one that changed: capital for the duration of a single transaction is available without collateral, because repayment is enforced within the same call.

Interpretation

The examination question shifts

Against logic errors, examining the code helps. Against this class, what helps is testing the rules for their assumptions: which observed quantity feeds a decision? How expensive would it be to move it for the duration of a block? And what happens if exactly that occurs? The third question is a modeling question, and it belongs in the analysis even when two audit reports exist.

Data

What the research places here

That assurance services in this field are themselves the object of empirical research, and that an audit's existence is a finding about the process, is recorded in EVD-2026-0005; that smart contracts have nameable limits around external data, in EVD-2026-0004. Both sources place mechanisms in general. They say nothing about any particular contract or incident, and they replace no audit.

Definitions

Economic exploit: An incident in which contracts execute as specified while the assumptions behind the rules do not hold.
Flash loan: An uncollateralized borrowing whose repayment is enforced within the same call — otherwise the whole call counts as not having happened.
Specification: The description of what a program should do — the yardstick an audit checks against.

Model

Which observed quantity feeds a decision of the contract?
How deep is the market that quantity comes from?
How large is the sum decided on its basis?
If line three exceeds line two, the finding must be recorded

Four questions that surface this class — All four figures are available before an incident. That is what separates this class from an undiscovered logic error.

Formulas

Rough attack arithmetic

worthwhile if achievable_gain > cost_of_moving + fees

achievable_gain: What the distorted quantity makes payable
cost_of_moving: Price impact and fees to shift the observed quantity briefly
fees: Execution costs including the in-transaction borrowing fee

Limit: The inequality describes when an attempt pays, not whether it is possible — and it knows the cost of moving only for the depth present today. If depth grows, the arithmetic flips without anything in the contract changing. It is a thinking tool, not a security assessment.

Worked example

A rule that is sensible taken on its own

Rule: collateral value follows the price in pool X
Depth in pool X: USD 1.4m in the tight band
Borrowable at a distorted price: up to USD 9m
Cost of moving the price for one block: roughly USD 300,000

Achievable gain well above the cost of moving: the rule takes its price from a market shallower than the sum it decides over. Every contract works correctly throughout.

No bug in the code, no rule broken — and a loss the size of the difference all the same.

Reading: The checkable finding is not “sloppy code” but: the price source is shallower than the sum depending on it. That is one number against another and can be established before an incident — so the question belongs in every examination of a collateralized position.

Retrieval

A logic error overlooked until now.An economic exploit: the rules held, their assumptions did not.A failure of the chain.

The number of audit reports available.The depth of the price source against the sum decided on its basis.The size of the bounty program.

Exercise on real data

Check the guiding questions for those needing an audit report and those answerable from market data.

Dimension 9: security →

Determine which of the figures named there says nothing at all about this class of failure.

Case study: audit present, question open →

Application

Write the examination line covering this class of failure for a collateralized position.

Related case studies

Institutional reading

Bank: Which of your controls would notice an incident in which no rule was broken?
Insurance: Under your wording, would an incident without a rule breach be a covered event?

Metrics in this lesson

Bad debt

Key takeaways

Economic exploits breach no specification, only the assumptions behind it.
The ingredients are a price-dependent rule, a movable source and short-term capital.
Price-source depth against the dependent sum is measurable in advance — and belongs in every examination.

Evidence

EVD-2026-0005
Review of Accounting Studies (Springer) — Decentralized Finance (DeFi) assurance: early evidence
EVD-2026-0004
Annual Review of Financial Economics — Smart Contracts and Decentralized Finance