What methods are used to reduce fluctuations in automated tests that sometimes produce false positives or false negatives?

To reduce fluctuations (instability) in automated tests, which sometimes produce false positives or false negatives, the following methods are used:

• Test idempotency — tests should be independent and not affect each other.
• Environment isolation — run tests in a controlled and clean environment to exclude external influences.
• Use of stubs and mocks — replace unstable external services with controlled mocks.
• Expectations and timeouts — properly use explicit waits instead of fixed delays to wait for the desired state.
• Retries — automatically rerun the test upon failure to filter out random failures.
• Parallel execution with caution — avoid race conditions and conflicts during parallel runs.
• Logging and monitoring — detailed logging for quick identification of instability causes.

Example: if a UI test sometimes fails because an element has not yet appeared, instead of Thread.sleep(5000), it is better to wait for the element to appear with a timeout. This reduces false positives caused by loading delays.