Citation Decay
A claim losing its source through repetition, until a number everyone cites has no traceable origin.
Reviewed July 16, 2026Stable
When not to use it
- As a way to dismiss a figure. Unsourced and false are different, and treating them as equivalent is its own error.
- Where the primary source exists and is simply not linked, which is sloppiness rather than decay.
- For genuinely contested figures where two parties each have methodology, which is a dispute rather than an orphaned claim.
Reach for something else instead
- Primary-source citation — reach the document and cite it, which is the whole answer and is more work.
- Explicit non-linking — name the document precisely and state that no stable URL resolves, which preserves checkability without pretending to a source.
- -
Read more on the blog
- Every chatbot query on earth is 2% of AI's powerTerritory 8 opens on the numbers behind AI's physical footprint, and on the arithmetic in the IEA's own report that almost nobody quotes.
- Same GPUs, same month, opposite depreciationTwo companies bought the same hardware, both were audited, and they reached opposite conclusions about how long it lasts. The difference flows straight into reported profit.
- Same company, same year, revenue figures 63% apartThe most quoted numbers in AI are run-rates from private companies, reported by outlets triangulating from leaks, and sources covering the same period differ by multiples.
- The water bottle was per 10 to 50 responsesThe most repeated environmental claim about AI dropped a qualifier from the study it cites. The concern it created is still justified, for different reasons than the number suggested.
Further reading
- Raji et al. (2021), AI and the Everything in the Whole Wide World Benchmark — how a specific measurement becomes a general claim through restatement.
- Lipton & Steinhardt (2018), Troubling Trends in Machine Learning Scholarship, arXiv:1807.03341 — the mechanisms by which claims outrun their evidence in a literature.
Primary sources, listed so you can check the claims on this page rather than take them on trust.
Where people go wrong
- Counting corroborations as evidence when they trace to one ancestor.
- Attributing a figure to the organisation it describes because it appears in coverage about them.
- Repeating a projection without its date and conditions, which is how forecasts harden into facts.
At a glance
FieldFoundations
Three formsorphaned statistic, projection hardening, attribution drift
Telloddly precise, identically phrased, no primary link
Interacts badly withprevalence-ranked retrieval
Countermeasurecite the document or state its absence
DifficultyBeginner → Intermediate
Where this sits
A starting point. Nothing needs to come before it.
0Levelstarting point
0Needs firstnothing
0Opens upnothing further
1Areastays here
Computed from the prerequisite graph, not assigned. How this works