Citation-version discipline (cite with a version/date stamp or don't cite it)
A single-week cluster of six independent 06-reference/research/ deep-research briefs (2026-10-12 through 2026-10-15) each closed the same way: a number the vault had been carrying as settled turned out to be stale, unversioned, or quietly re-derived from a different denominator than the one originally cited.
- [[2026-10-13-uspto-pendency-and-itu-extension-fee-2026]] killed a carried "8-12 months to register" figure — current USPTO pendency is 4.3/10.4 months, and the old figure had no date stamp attached to it in the vault, so nothing flagged it as decaying.
- [[2026-10-15-nber-w35275-writing-vs-shipping-code]] found the vault's own "741%/20%" productivity pairing was citing an earlier, unversioned preprint revision — the current NBER working-paper version reports 958%/20%, and the metric itself is lines-changed, not "code generated," a unit slippage that compounded the staleness.
- [[2026-10-14-micron-cip-contamination-unwind]] reframed a single carried threshold (
shell_share) as a band (29-57%) once the primary filing was re-pulled, because the original citation had collapsed a range to a point estimate. - The three Virta/GLP-1 research siblings — [[2026-10-12-glp1-discontinuation-cohort-past-18-months]], [[2026-10-12-lookahead-direct-remission-itt-vs-completer]], [[2026-10-12-virta-drcp-medication-effect-score-itt-reconstruction]] — each independently found that a headline clinical number (an 18-month durability claim, a remission-rate comparator, a medication-reduction percentage) only holds at one specific denominator (completer vs. ITT, enrolled vs. randomized), and the vault's prior citations had dropped the denominator along with the number.
The common failure shape: a number gets cited once, correctly, against a primary source at a point in time — but the citation is stored as a bare fact ("8-12 months," "741%," "18-month durability") rather than as a fact-plus-provenance ("8-12 months, per USPTO FY2025 pendency report, as of [date]"). Once the provenance is dropped, nothing in the vault can tell the difference between a number that is still true and one that has quietly decayed, been revised upstream, or only ever held at a denominator nobody wrote down. The research brief that eventually re-pulls the primary source does the correction, but the correction is only possible because someone bothered to re-check — there is no structural signal that a given number is due for a recheck.
Why this is in the vault
Six deep-research briefs hitting the identical failure mode independently, in the same five-day window, across completely unrelated domains (trademark law, AI-coding productivity, semiconductor supply chains, metabolic-health RCTs) is a pattern, not a coincidence — it is what happens when a vault accumulates numeric claims faster than it re-verifies them. /compile-vault and /deep-research are exactly the mechanisms that catch this, but only when a brief happens to touch the same topic again; there is no proactive "this number is N months old, re-check it" signal. Naming the failure mode is the first step toward building one (a verified_as_of frontmatter field, or a standing /deep-research sweep over citations older than some threshold, are the natural next builds — not undertaken here, flagged for whoever picks this up).
Mapping against Ray Data Co
- Research pipeline (
/deep-research,06-reference/research/): every brief in this cluster already does the correction work — the gap is upstream, in how a number gets written down in the first place and how long it sits before anyone re-checks it. The fix is procedural: cite with the source name and the date/version of that source, not just the figure, so a future brief can tell at a glance whether a number is primary-sourced-and-current or inherited-and-aging. - phData / credibility content: several of these corrected figures (NBER productivity study, USPTO pendency) are exactly the kind of number that could end up in a Sanity Check piece or a credibility-building post — citing the decayed version publicly would be a real (if small) reputational cost. This is the concrete stake behind the discipline.
- Investing/acquisitions research: the Micron CIP-contamination case shows the same decay risk applies to any thesis-supporting number pulled from a filing — a point estimate cited without its original range is a silent landmine in a later investment memo.
- Health/longevity project: the three Virta/GLP-1 briefs feed the vault's own prior longevity thesis notes (e.g. [[2026-05-22-tim-ferriss-sami-inkinen-virta-t2d-rowing]]); this cluster is the concrete evidence that denominator-tier discipline, not just directional correctness, is required before a clinical figure gets reused in planning.