Four-Eyes Principle

Highly secure systems often implement the four-eyes principle — a control mechanism requiring two people to approve critical changes before implementation.

The idea is much older than computers (banks have long required two signatures on large checks), and it guards against two different problems: honest mistakes that a second person would catch, and a single insider who decides to go rogue. In software teams, the principle usually takes the form of:

  • Required code review, so no change reaches the main branch without a second engineer approving it.
  • Approval steps in the deployment pipeline for releases to production.
  • Two-person rules for sensitive operations, such as deleting customer data or issuing refunds above a set amount.

The control only works if the second person actually looks. A reviewer who approves every request without reading it adds a delay but no security. The principle also works best alongside the principle of least privilege, since it is much easier to make sure a second person checks every change when only a few people can make changes at all.

Further Reading