A Terraform plan can succeed and still deploy the wrong thing.
The issue is not syntax. It is readability.
When expressions are written inline with multiple transformations, the final value is correct from Terraform’s perspective, but the logic that produced it is hidden from you. That forces you to mentally execute the expression during review.
That is where mistakes pass through.
Locals fix this by separating each transformation into a named step.
Each value becomes inspectable.
Each decision becomes visible before apply.
Terraform only evaluates outcomes.
You are responsible for validating intent.
Readable expressions create a boundary where you can catch errors before they reach AWS.