Modeling Fidelity Versus Deadline
A city transit authority must publish a target-state architecture for a fare-collection overhaul before a funding submission in 10 weeks. The current-state documentation is 3 years old and known to be incomplete: two bus depots changed ticketing vendors and one rail line was added. The architecture team has 4 people, two of whom are part-time. Stakeholders include operations, a payments vendor, the accessibility office, and a funding body that expects quantified benefits. The team proposes spending 6 weeks building a fully validated current-state model before any target-state work begins.
1. Evaluate the proposed 6-week current-state-first plan against the planning constraints.
Read the worked answer
The plan is likely misallocated. Reasoning:
- Time budget: 6 of 10 weeks on current state leaves 4 weeks for target state, stakeholder agreement, and benefit quantification-too little for four stakeholder groups.
- Purpose mismatch: the deliverable is a target-state architecture for funding. Current-state fidelity matters only to the extent it constrains the target or supports benefit estimates.
- Known gaps are bounded: two depot vendor changes and one added rail line are enumerable; full validation of everything is over-scoping.
- Part-time staffing means validation will be slower than assumed.
Recommend a scoped current-state effort (e.g., 2-3 weeks) focused on fare-relevant elements, running in parallel with target-state option development. The answer should tie effort to decision value, not completeness for its own sake.
2. Recommend how to handle the known stale and incomplete current-state information within the timeline.
Read the worked answer
A workable approach:
- Prioritize by impact: model only current-state elements that affect fare collection (ticketing systems at the two changed depots, the new rail line, payment interfaces).
- Record assumptions explicitly: mark unvalidated elements with a confidence/assumption register so the funding body sees known uncertainty.
- Targeted validation: confirm the enumerated changes with the vendors and operations rather than re-validating the whole estate.
- Parallelize: start target-state options immediately, refining as current-state facts land.
- Accessibility and payments stakeholders review early because their constraints shape options.
The answer should show that incomplete current state is managed through scoping and explicit assumptions, not by pretending it is complete or by blocking on it.
3. The funding body requires quantified benefits. How does modeling approach affect the credibility of those numbers?
Read the worked answer
Benefit credibility depends on the link between modeled elements and the quantities claimed. If the current-state model is stale, any baseline (e.g., current fare leakage, transaction cost) inherits that error, and the funding body may discount the whole submission.
Recommendations:
- Tie each benefit to a specific modeled element and its stated confidence.
- Present benefits as ranges where baseline data is uncertain, with the assumption register cited.
- Validate the few baseline figures that drive most of the benefit (likely transaction volume and leakage) rather than all figures.
- Show traceability from target-state change to benefit mechanism to baseline source.
The answer should make clear that selective, well-sourced baselines beat comprehensive but unverified modeling for funding credibility.