Research

Research

How we measure the design factory: the methods, the data and every result we publish, including the runs we lose.

• Benchmark

  Nordform Trueline: does a design factory produce what it claims?

  Briefs in, designs out. Trueline grades whether each answer is the one the brief asked for, whether it stops when it should, whether independent readers open the files, and what it costs.

  Corpus plumb-v1 · last run 2026-10-21 on plumb-v1 · 3 published runs

• Working paper

  Design is what the project waits for

  Eelon Lappalainen argues that in industrial investment projects engineering design is less a cost than the constraint that sets the project's pace, so design time should be valued at the project's cost of delay, and proposes three measures to test it.

  Working paper, version 0.1 · 21 October 2026

How we publish

• Every result names its corpus version and its date.
• Since the corpus was frozen on 20 October 2026, every run that covers the held-out cases is published, whatever it shows. A result is never removed; a correction is added beside it.
• The cases, graders and results are open: cases and results under CC BY 4.0, code under Apache-2.0.