--- name: A Value You Did Not Measure Still Has A Source slug: a-value-you-did-not-measure-still-has-a-source category: Writing description: A Value You Did Not Measure Still Has A Source explains how to handle deliverables this run cannot measure or produce, using attributed published values instead of fabricating or omitting them. It covers where to place cited values and how to compare them to your own results. github: "https://github.com/tangxiangru/AutoR/tree/main/src/skills/a-value-you-did-not-measure-still-has-a-source" language: Python stars: 804 forks: 25 install: "npx degit https://github.com/tangxiangru/AutoR/tree/main/src/skills/a-value-you-did-not-measure-still-has-a-source ~/.claude/skills/a-value-you-did-not-measure-still-has-a-source" installs_to: ~/.claude/skills/a-value-you-did-not-measure-still-has-a-source source_path: src/skills/a-value-you-did-not-measure-still-has-a-source/SKILL.md collection_size: 25 category_size: 1012 collection_url: "https://dirskills.com/collections/tangxiangru/AutoR" added: 2026-08-22T05:22:23.026Z last_synced: 2026-08-22T05:22:23.026Z canonical_url: "https://dirskills.com/skills/a-value-you-did-not-measure-still-has-a-source" --- # A Value You Did Not Measure Still Has A Source A Value You Did Not Measure Still Has A Source explains how to handle deliverables this run cannot measure or produce, using attributed published values instead of fabricating or omitting them. It covers where to place cited values and how to compare them to your own results. **Install:** ```bash npx degit https://github.com/tangxiangru/AutoR/tree/main/src/skills/a-value-you-did-not-measure-still-has-a-source ~/.claude/skills/a-value-you-did-not-measure-still-has-a-source ``` ## README # Not measuring something is not the same as having nothing to say about it Some tasks name a deliverable this run cannot produce. The study asks for a synthesised polymer and a DSC trace; you have no bench. It asks for a run on hardware you do not have, or a benchmark behind a licence, or an experiment that takes six weeks. There are three things you can do, and they are not close in value: 1. **Omit it.** The section does not exist. To a reader this is indistinguishable from a run that forgot, and it is scored as absence. 2. **Name the gap.** One paragraph: this was asked for, it could not be done, here is why. Honest, and better than silence. 3. **Report the field's own answer, attributed, and position your work against it.** The value exists — in the source study, in a reference database, in a handbook. Give it, say whose it is, and show what your work does and does not establish about it. The third is what a real paper does. A methods section that needs a melting point it did not measure cites one. A benchmark table with a row you could not run gives the published figure with a citation and a footnote. This is ordinary scholarly practice, and it is the difference between a section that says nothing and a section that positions the run's contribution in the literature. ## Fabrication and citation are different acts The rule you are working under is that no number in the report may be invented, estimated, or narrated into existence: every quantity traces to something. That rule is correct and this does not weaken it. **A published value has a source; it is traceable; it is not an invention.** What makes the difference is entirely in how it is carried: - **Attributed.** "Zheng et al. measured DSC 317 K and TMA 311 K on their synthesised candidate (ref. 4, Fig. 5)." - **Labelled where it appears.** In a figure, the external value is a separately styled band or marker with the source in the legend — never a series that reads as one of yours. In a table, a `source` column. - **Never counted as validation of your own pipeline.** The published number is the target your work is compared against, not evidence that your work is right. A number carried that way is a citation. The same number carried as your own measurement is a fabrication. The line is bright and it is about labelling. ## Plotting the value is not enough; you have to stand next to it The near-miss is worth naming because it looks like the right answer. A run puts the published band on a figure, and then captions it "carried as an external reference, and never as a validation of this run" — and the section still reads as a refusal, because nothing of the run's own is placed against the band. A reader sees a number that belongs to someone else and no claim. The version that works puts the run's own quantity in the same frame and states the distance: "our pipeline predicts 355 K against the 311-317 K they measured — 0.8σ of our uncertainty budget." Same published number, same honesty about who measured what, and now there is a result in the sentence. The disclaimer is still correct and still belongs there. It goes *after* the comparison, as a qualification of it, not instead of it. ## Then say what your work does establish about it This is the part that turns a citation into a result. You could not synthesise the candidate — but you designed it, and you have a predicted property with an uncertainty. Put the prediction and the published measurement in the same figure and state the deviation. You could not run the licensed benchmark — but you have your metric on the open subset, and the source reports both, so the offset between them is estimable. The section then reads: this is what was asked, this is what the field has measured, this is what this run predicts, this is the gap between them, and this is what would close it. That is a contribution. "Not attempted" is not. ## Where it goes In the section a reader looks in for the answer — under the heading the task's own words would send them to — not in Limitations. Limitations gets a cross-reference. A deliverable answered only in Limitations has been answered in the one place a reader goes to find out what the run failed at. ## The check For each named deliverable the run could not produce: is there a section under a heading a reader would look for, containing the field's value with its source, this run's nearest evidence, and the distance between them? If the answer is "it is mentioned in Limitations", it is not done. See also `cover-what-the-task-named` for enumerating deliverables in the first place, and `citation-discipline` for how the source is recorded once you cite it.