Why good work goes unbuilt
The most expensive projects are the ones that never happened. On paper, they cost nothing. About 70% of manufacturers were stuck in “pilot purgatory” (McKinsey, 2018), and 91.5% of big projects run over budget, over schedule, or both (Flyvbjerg). White paper #1: write the test before you build, then let the evidence decide. Proof over promises.
Escherbridge · · 10 pages · US Letter

Summary
Good work rarely dies because someone lacked the invention. It dies in three quieter ways: pilots that work and then go nowhere, projects that do happen at a price nobody agreed to, and the people with a problem, the people with the tools, and a place to test an answer simply never meeting. This paper walks through the research on each, from McKinsey’s “pilot purgatory” and Gartner’s abandoned AI proofs of concept to Bent Flyvbjerg’s “iron law of megaprojects” and the end of the TradeLens shipping platform. It finds one shared flaw: the test was never agreed, so the evidence could never decide. It then sets out a different order of operations, the PrimusNeo clearing house plus innovation lab, where the measures are set before anything is built and “stop” counts as an allowed result. It describes the underlying stack in plain words and uses Tie Town [TX], one co-op site in McDade, Texas, as a small, honest example. It ends with four questions to answer before building anything.
What’s inside
- The loss nobody counts
- Pilot purgatory
- The overrun
- The missed meeting
- A different order of operations
- The plumbing, briefly
- Tie Town [TX]
- Before you build
- Sources & notes
10 pages · US Letter
Download PDFDirect file. Nothing to fill in, and nothing is stored because you clicked it.
Talk to us.
The paper ends with four questions. If they already have answers, you may not need this. If they do not, the contact page is the same form the rest of the site uses.
Talk to usPrefer the notes to a form? The feed carries the blog, the newsroom, and these papers.
Related
The paper describes the stack in plain words. These are the pages where the site describes the same layers, including the roadmap, where the status is written down.
The stack
Roadmap
- AMP on the roadmapShipped. Tie Town [TX] is the first live site.
- NEOS on the roadmapShipped. Tie Town [TX] is the first live site. Not every skill is claimed as in use.
- AZOA on the roadmapIn alpha. Consent is in. Real-value bridges stay off.
- Fractal Engine on the roadmapIn alpha. The limits are part of the description.
- Elevate Impact on the roadmapLive as research. The join to the four layers is in build.
- RoadmapThe whole list. Shipped, in alpha, building, or exploring. No dates.
Notes
- A digital twin shouldn’t live in someone else’s cloudThe pilot that becomes the product because leaving was never part of the design.
- Shared ownership, unshared minutesA co-op needs a record that outlives the people who were in the room. The paper’s example is one site of that kind.
- Alpha is a status, not a complimentWhat “not finished” looks like when it is written down. The paper does the same for the stack.