The Computing Series

What Goes Wrong

The most common engineering economics failure is the invisible cost of maintenance. Teams invest in building capabilities, account for the infrastructure cost, and do not model the ongoing engineering overhead — the time spent on incidents, the time spent on updates, the cognitive load of understanding the system in a high-turnover environment. FM3 (Unbounded Resource Growth) in the maintenance dimension: every capability added is a permanent maintenance obligation that does not appear in the investment proposal.

The second failure is cost of delay in reverse: prioritising work by engineering preference or backlog age rather than economic urgency. A feature with a ten-thousand-euro monthly cost of delay that has been in the backlog for six months has cost sixty thousand euros of delay. A feature with a one-thousand-euro monthly cost of delay that ships this sprint costs one thousand euros of delay. The backlog ordering by engineering preference has cost the organisation fifty-nine thousand euros.

FM6 (Hotspotting) at the economic level: infrastructure costs that are concentrated in a small component — a single database, a single service, a single data transfer path — make the cost curve non-linear. When the hot component reaches its capacity ceiling, the cost to continue scaling jumps discontinuously. Engineering organisations that do not model their cost hot spots are surprised by cliff-edge cost events that could have been predicted from the cost model.

Concept: Engineering Economics Thread: T12 (Tradeoffs) ← technical property → economic consequence → investment decision Core Idea: Engineering decisions are economic decisions; cost of delay, total cost of ownership, and scaling cost curves are the vocabulary that makes technical investments legible to non-technical stakeholders. Tradeoff: AT2 — latency (higher infrastructure cost per unit) vs throughput (lower cost per unit at scale) Failure Mode: FM3 — unbounded resource growth; maintenance overhead of capabilities not modelled in investment proposals Signal: When a board or executive asks “is this worth it?” and engineering cannot answer with numbers — the investment was not framed economically; stop and produce the three-number structure before proceeding Maps to: Reference Book, Frameworks 4, 9

Read in the book →