A system can define roles such as requester, reviewer and administrator, each with an explicit set of allowed operations. Assigning identities to those roles makes recurring permission decisions easier to manage and can separate sensitive responsibilities such as requesting and approving a refund.
The trusted server must enforce the permission decision using verified identity and current policy. Hiding a button does not prevent an unauthorized request, and a role may still need additional checks on the specific record, amount or action context.
ELI5
Imagine different staff badges opening different doors. A receptionist's badge and a manager's badge have different permissions because their responsibilities differ.
For example, a support agent can prepare a refund request while a manager approves it. The server checks both identities and the allowed refund conditions, rather than trusting which buttons the browser displayed.
Is hiding a control enough to enforce a role?
No. The server must independently check whether the verified caller is allowed to perform the requested operation.
Can one role authorize every action on every record?
Only if that is the intended policy. Sensitive workflows often need narrower record-level or contextual checks in addition to roles.

