A deterministic test controls the sources of uncertainty

First, build a mental model

A test is an experiment: arrange controlled inputs, perform behavior and compare the observable result with a requirement. Dependency injection makes a source of uncertainty explicit so a focused test can replace it.

1. A unit test isolates a rule; an integration test checks real collaborators and boundaries. Both are needed for different failure modes.
2. Inject a clock rather than reading current time inside every rule; boundary cases then become reproducible.
3. A test double can control a dependency, but cannot prove the real implementation's database or network behavior.
4. Use the project's configured test framework and `dotnet test`; a green command matters only if the intended tests actually ran and the asserted contract is useful.

static bool IsExpired(DateTimeOffset expiresAt, IClock clock)
    => clock.UtcNow >= expiresAt;

var boundary = DateTimeOffset.Parse("2026-01-01T00:00:00Z");
IClock clock = new FixedClock(boundary);
Console.WriteLine(IsExpired(boundary, clock));

interface IClock { DateTimeOffset UtcNow { get; } }
sealed class FixedClock : IClock
{
    public DateTimeOffset UtcNow { get; }
    public FixedClock(DateTimeOffset now) { UtcNow = now; }
}
// In your test framework, assert before / exactly-at / after expiry.
// This page does not run dotnet test or claim a test passed.

1. Requirement

   The example contract says equality with the expiry instant counts as expired.

2. Control

   FixedClock supplies an exact instant, so wall-clock timing cannot make the result flaky.

3. Observe

   The method returns true at the boundary. Neighboring cases should be tested separately.