Exercise 3: The Risk Reduction Needed for a Reversible Decision — Possible Solution
====================================================================

METHOD
------------------------------
Solving for ADR_RISK_REDUCTION where with_adr_cost == without_adr_cost,
at reversal_cost=1:

    2 + 0.3*rr*1 = 0.3*1
    2 + 0.3*rr = 0.3
    0.3*rr = -1.7
    rr = -5.67

RESULT
------------------------------
    Required ADR_RISK_REDUCTION for break-even: -5.67
    (a valid risk reduction is between 0 and 1 - this value is impossible)

    Best possible case (ADR eliminates 100% of risk, risk_reduction=0):
      with_adr=2.00, without_adr=0.30
      Even in the best possible case, ADR is still worse: True

No valid value of ADR_RISK_REDUCTION (which must fall between 0 and 1,
since it represents a genuine risk reduction, not risk elimination
beyond 100%) makes writing an ADR worthwhile for this specific
reversible decision - not even the impossible best case where the ADR
eliminates the wrong-decision risk entirely.

WHY THIS WORKS AS AN ANSWER
------------------------------
This is a genuinely striking, honest result: for a decision this cheap
to reverse (1 hour), the ADR's own fixed 2-hour writing cost alone
exceeds the ENTIRE possible expected cost of just getting it wrong and
fixing it (0.3 hours) - even before any question of how much the ADR
actually reduces risk. No amount of "the ADR makes us think more
carefully" can close a gap that large, because the problem isn't the
ADR's own effectiveness, it's that the stakes were never big enough to
justify its fixed cost in the first place. This is the clearest
possible confirmation of the chapter's own core rule: for a
sufficiently reversible decision, the question isn't "how good is our
ADR process" - it's "should we be writing one at all."