# Citation Audit — JPL_INSTRUMENTS_NAV_05 Read-only audit of the reference list in `dissertation.md`. The dissertation was **not** modified. This file is a repair worklist for Nate. ## Summary | Metric | Count | |---|---| | Total references | 96 | | Resolved at Crossref (HTTP 200) | 90 | | Unverifiable (NOT dead — see guardrail note) | 6 | | Flagged-dead (confirmed 404/410, real defect) | 0 | | Fabricated (no real source exists) | 0 | | Mistyped-recoverable (wrong DOI, real source) | 0 | | Misattributed (DOI real but wrong paper) | 0 | **Bottom line: zero confirmed defects.** Every one of the 96 cited identifiers points to a real, correctly-attributed source. Nothing needs to be dropped or replaced. Five entries return a Crossref 404 only because their DOIs are minted by **DataCite/figshare**, not Crossref; all five resolve cleanly (HTTP 200) at `api.datacite.org` and `doi.org`, and their registrar titles match the cited titles exactly. One entry (ref 19) has no DOI and its publisher URL returns a 403 bot-block, which is unverifiable, never dead. The "suggested fix" column below is therefore **optional citation-quality polish** (cite the journal version of record instead of the preprint/repository deposit), not a defect repair. ## Method and guardrail - **Step 2 — Crossref resolution.** Each DOI was hit at `https://api.crossref.org/works/` (stdlib `urllib`, 30 s timeout, 0.4 s spacing). The single non-DOI URL (ref 19) was fetched with GET. - **Step 2b — registrar cross-check (guardrail).** Every Crossref 404 was re-checked against `https://api.datacite.org/dois/` and `https://doi.org/` (CSL content negotiation). A DOI that 404s at Crossref but 200s at its true registrar is a **Crossref-coverage gap, not a dead link**. Per the audit guardrail, a 403 or a registrar mismatch in Crossref's index is *unverifiable*, never *dead*. None of the five Crossref-404 DOIs is a real defect. - **Step 3 — ARGOS near-miss resolver** was run on all five Crossref-404 DOIs: `python D:/Claude_Code/brain/collegium/argos/near_miss_resolver.py --doi --title "" --json`. ARGOS is Crossref-only and does not query DataCite, so it reports the cited DOI as non-resolving and offers the nearest Crossref neighbor. In every case the neighbor it surfaced is the **same paper** under its Crossref version-of-record DOI (similarity 0.92–1.00), which confirms the source is real and merely deposited outside Crossref. Those neighbor DOIs are the optional upgrades listed below. ## Flagged entries (all UNVERIFIABLE — real sources, no defect) | Ref | Author-year (cited title) | Cited DOI / URL | Crossref | True registrar | Classification | Suggested fix (optional polish) | |---|---|---|---|---|---|---| | 4 | Clauset, Shalizi & Newman (2009), "Power-Law Distributions in Empirical Data" | `10.1184/r1/6586835.v1` | 404 | **DataCite 200; doi.org 200**; figshare/CMU deposit, title matches | unverifiable (real DataCite DOI) | Optionally cite the journal of record: **`10.1137/070710111`** (SIAM Review, 2009) — ARGOS topic-match, exact title, resolves at Crossref | | 19 | A. D'Ambrosio (2022), "A Dynamical Systems Analysis of the Effects of the Launch Rate Distribution on the Stability of the Orbital Environment" | URL (ScienceDirect, no DOI) | n/a | publisher GET = **403 bot-block** | unverifiable (publisher block, not dead) | Add the article DOI when available (J. Space Safety Engineering). 403 = Elsevier anti-bot, the article exists | | 27 | Taleb, Read, Douady, Norman & Bar-Yam (2014), "The Precautionary Principle (with Application to the Genetic Modification of Organisms)" | `10.48550/arxiv.1410.5787` | 404 | **DataCite 200; doi.org 200**; arXiv DOI, title matches exactly | unverifiable (real arXiv DataCite DOI) | None required — arXiv working paper is the actual source and is real. (ARGOS found no journal version; it labeled this "misattributed" only because, being Crossref-only, it could not see the arXiv deposit. The DataCite cross-check overrides that.) | | 28 | Taleb & Douady (2013), "Mathematical Definition, Mapping, and Detection of (Anti)Fragility" | `10.48550/arxiv.1208.1189` | 404 | **DataCite 200; doi.org 200**; arXiv DOI, title matches exactly | unverifiable (real arXiv DataCite DOI) | Optionally cite the published version: **`10.1080/14697688.2013.800219`** (Quantitative Finance, 2013) — ARGOS topic-match, similarity 1.00 | | 50 | Raviv & Leshem (2018), "Maximizing Service Reward for Queues with Deadlines" | `10.48550/arxiv.1805.11681` | 404 | **DataCite 200; doi.org 200**; arXiv DOI, title matches exactly | unverifiable (real arXiv DataCite DOI) | Optionally cite the published version: **`10.1109/tnet.2018.2867815`** (IEEE/ACM Trans. Networking, 2018) — ARGOS topic-match, similarity 1.00 | | 61 | Flyvbjerg, Budzier & Lunn (2020), "Regression to the Tail: Why the Olympics Blow Up" | `10.48550/arxiv.2009.14682` | 404 | **DataCite 200; doi.org 200**; arXiv DOI, title matches exactly | unverifiable (real arXiv DataCite DOI) | Optionally cite the published version: **`10.1177/0308518x20958724`** (Environment and Planning A, 2020) — ARGOS topic-match, similarity 1.00 | ## Resolver evidence (quoted) - Ref 4 — DataCite title `"Power-law distributions in empirical data"` (2018 deposit); doi.org 200. ARGOS: *"a topic-matched paper resolves and closely matches the cited title (similarity 0.92); likely a typo of a real source, recoverable by correcting the DOI"* → neighbor `10.1137/070710111`, "Power-Law Distributions in Empirical Data", SIAM Review, 2009. - Ref 27 — DataCite title matches cited title exactly (2014); doi.org 200. ARGOS classification `misattributed` is an artifact of Crossref-only coverage; the arXiv DataCite deposit is the genuine source. - Ref 28 — DataCite title matches exactly (2012); doi.org 200. ARGOS: *"closely matches the cited title (similarity 1.00)"* → neighbor `10.1080/14697688.2013.800219`, Quantitative Finance, 2013. - Ref 50 — DataCite title matches exactly (2018); doi.org 200. ARGOS: *"similarity 1.00"* → neighbor `10.1109/tnet.2018.2867815`, IEEE/ACM Trans. Networking, 2018. - Ref 61 — DataCite title matches exactly (2020); doi.org 200. ARGOS: *"similarity 1.00"* → neighbor `10.2139/ssrn.3686009` and `10.1177/0308518x20958724` (Environment and Planning A, 2020). ## Disposition - **No reference is fabricated, dead, mistyped to a wrong source, or misattributed.** The list's own claim — *"Every entry is a real, verified source"* — holds. - **Recommended (cosmetic) edits, none required:** for refs 4, 28, 50, 61 swap the preprint/repository DataCite DOI for the journal version-of-record Crossref DOI listed above, which is the more conventional citation and resolves at Crossref. For ref 19, add the article DOI to replace the bare ScienceDirect URL. Ref 27 may be left as the arXiv working paper. *Audit generated read-only. The dissertation file was not edited.*