Why is it a mistake to treat Ansible like a scripting language?

Question

Accepted Answer

Ansible is a desired-state engine: you describe what the end state should be, and modules make only the necessary changes. Writing step-by-step commands throws away idempotency and readability.

Imperative (script)
Declarative (Ansible)

You write
How: run this command, then that one
What: package X present, service Y started

Running it twice
May repeat actions, append duplicates, fail
Changes nothing if already correct

Reading it
You have to simulate the steps in your head
The file states the target state directly

A playbook full of shell: tasks is really a shell script in YAML form: it runs every command each time, reports "changed" constantly, and hides the intended end state. Use modules that express state (package, service, template, file) and Ansible works as designed.
Go deeper:

Good Practices for Ansible (Red Hat Communities of Practice) — why declarative beats imperative in playbooks, with concrete do/don't examples.
 Declarative programming (Wikipedia) — the general "what, not how" paradigm compared with imperative code.

	Imperative (script)	Declarative (Ansible)
You write	How: run this command, then that one	What: package X present, service Y started
Running it twice	May repeat actions, append duplicates, fail	Changes nothing if already correct
Reading it	You have to simulate the steps in your head	The file states the target state directly