Challenge 2: Identify the Manual-Process Failure Mode — Possible Solution
====================================================================

THE SCENARIO
-------------
A team's release process requires a developer to manually SSH into a
production server and run a deploy script.

TWO DISTINCT FAILURE MODES
-------------------------------
1. Forgotten or skipped steps under deadline pressure

   The chapter explicitly names this: developers "skip [tests] 'just
   this once' under deadline pressure" and manual processes create
   "forgotten steps exactly when there's least time to catch them."
   Manually SSHing in and running a script depends entirely on the
   developer REMEMBERING every required step in the correct order (e.g.
   pulling the latest code, running any needed migration first, then
   the deploy script, then restarting the right service) — under time
   pressure, a step can easily be skipped, run out of order, or run
   against the wrong environment entirely, with nothing structurally
   preventing that mistake.

2. Environment inconsistency ("works on my machine")

   The chapter names this directly too: manual processes create
   "inconsistency between environments." If different developers SSH in
   and run the deploy script slightly differently (different working
   directory, different assumed environment variables already set in
   their own shell session, a slightly different version of some tool
   installed locally), the deployment's actual behavior can vary
   depending on WHO ran it and WHAT their personal setup happened to
   be at that moment — rather than being a fully reproducible,
   identical process every single time, regardless of who runs it.

WHY THIS WORKS AS AN ANSWER
------------------------------
Both failure modes trace back to the same root cause the chapter
identifies: "every one of these steps is a place for human error." SSH
access plus a manual script execution is a textbook version of exactly
the "manually copy files to a server... manually restart the service"
pattern from this chapter's own opening description of the
pre-CI/CD release process — the specific mechanism (SSH + a script,
rather than manually copying files) doesn't change the underlying
category of risk; it's still a human-dependent step with no automated
safeguard ensuring it happens correctly and identically every time.