Tensions in Practice: One decisive score in tension with visible tradeoffs¶
Three invented delivery plans
Consider three plans: A is cheap and slow, B is costly and fast, and C is costly and slow. Here “cheap” and “costly” mean two fixed cost levels, and “fast” and “slow” mean two fixed delivery times. Making cost the sole objective selects A. Keeping both aims visible rules out C—another plan is no worse on either aim and better on one—but leaves A and B incomparable until a preference is supplied.
Reach a definite choice
Use an explicit priority to select a plan for action.
Keep conflicting aims visible
Avoid presenting a preference between cost and speed as if mathematics supplied it for free.
Why these aims pull against each other
A single score can give a definite answer by embedding a value judgment. Keeping the objectives separate exposes the tradeoff but may leave several undominated choices.
Choose an arrangement to see what changes and what remains difficult.
A qualitative three-plan comparison. Arrows in the second arrangement mean “dominates on the stated aims,” not movement or causation.
What this choice protects
What it costs
When it fits
Compare the arrangements
Give cost all the weight
Declare cost the sole objective for this decision and select A from the three plans.
- What it protects
- The stated priority gives a clear answer without pretending the choice is preference-free.
- What it costs
- B’s faster delivery cannot affect this objective, even if it would matter to a different decision-maker.
- When it fits
- Fits when lower cost really is the operative priority and every plan already satisfies any hard delivery requirement.
Illustration note: Giving cost all the weight is a limiting, transparent scalar objective. It illustrates the weighting choice rather than recommending that speed be ignored.
Keep cost and speed separate
Compare the plans on both dimensions. Exclude C, while retaining A and B as different tradeoffs.
- What it protects
- A and B remain visible rather than being silently ordered by an unstated weight.
- What it costs
- The result is not a final action. Someone must still decide what the cost of faster delivery is worth.
- When it fits
- Fits when the preference between aims remains open. Any hard constraints should first remove infeasible plans.
Illustration note: The two dominance edges are a complete qualitative comparison of this invented three-plan set. They do not estimate a real frontier.
What this illustration does—and does not—establish
Optimization: Single vs Multiple Objectives supplies the scalarization versus Pareto-choice tension. “Undominated” means no available alternative is at least as good on every stated objective and better on one; the diagram exhibits that relation directly.
- Cost and time labels denote fixed equal levels across plans in this toy; otherwise the displayed dominance claims would require more information.
- Being undominated does not certify feasibility, fairness, desirability, or the presence of every important objective.
- An explicit weighted score can be appropriate. The problem is an unexamined weighting presented as a preference-free best answer.
Source entries
Optimization
This source passage supplies the contextual tension. The concrete arrangements and schematic examples are editorial illustrations, not measured findings.
Single vs Multiple Objectives
- Structural tension: Real decisions usually involve multiple objectives that cannot be combined into a single cardinal score without value judgments. Collapsing them to a scalar hides the trade-offs; keeping them separate requires Pareto-style reasoning and a downstream choice among non-dominated alternatives. The choice of weights is itself the most consequential decision and is often made without conscious examination.
The source operation
Every optimization problem expresses as a triplet — *what to vary, what to value, what to respect* — extended by a fourth element specifying *the sense in which best is meant*: (1) decision variables or choice set over which the search ranges, (2) an objective function assigning a value to each candidate, (3) constraints that any admissible candidate must satisfy, and (4) the operative notion of optimality — exact global, ε-approximate, local, Pareto in multi-objective settings, or stochastic in expectation.