Temporal-Patent Fusion
A “mosaic theory” view that places two independent datasets on one timeline: the density of UFOSINT-catalogued sightings, and the activity of U.S. patents placed under a 35 USC §181 secrecy order. The question it explores is narrow and testable: does state-suppressed capability development co-move in time with phenomenon reporting? The page fuses four axes - temporal, spatial, semantic, and secrecy - each as its own panel below, building from the dual-signal timeline to the directional lead/lag analysis.
Read this first - what this tool is, and is not
- Hypothesis generation, never proof. Co-occurring activity is a lead worth investigating, not evidence of any causal claim. A correlation here is consistent with multiple explanations, mundane and otherwise.
- Pervasive shared causes.Both series are inflated by the Cold War aerospace boom, population and reporting-infrastructure growth, and military-region concentration of both R&D and sightings. Raw co-movement is expected and does not imply a connection.
- The disclosure paradox.A secrecy order deliberately obscures a patent’s assignee and location, so the most interesting patents are precisely the ones hardest to place in space. The robust fusion dimension is time (filing and imposition dates survive obfuscation); spatial fusion, added later, is supplementary.
- The statistics are honest about themselves. The lead/lag analysis below uses a circular-shift permutation null and a family-wise threshold that corrects for testing many lags at once. Where a pattern does not clear that threshold, the tool says so plainly rather than implying a link.
Does one signal lead the other?
The timeline above shows the two series side by side; this asks the harder, falsifiable question. By sliding the patent series forward and backward against sightings and measuring correlation at each lag - then comparing it to a null built by randomly rotating one series thousands of times - we can tell whether any lead/lag relationship is real or just what noise with the same internal structure would produce anyway.
Which capability classes are the strongest leads?
The Mosaic Score collapses the analysis to a single ranked list of patent technology classes. It blends how heavily each class was placed under secrecy, how long it stayed suppressed, and how strongly its secrecy timing co-moves with sightings - then applies a false-discovery-rate correction so the table can say honestly when a lead does not hold up. This is the temporal version; spatial and semantic terms are deferred.
Where are the facilities?
The temporal analysis above is the robust dimension; this is the supplementary spatial one. It places 1,394 typed defense and research facilities - military bases by service branch and research sites by capability category - on one map, filterable by their operational window, with the community sighting field and DECUR documented cases overlaid so facility-vs-sighting geography is visible at a glance. The sighting layer shows only the geocodable subset of reports (~385.7k of 618.3k total), so it is a spatial sample biased toward populated, well-named places. The map shows the geography; the panel below turns it into a measurement, testing which facility types draw sightings more tightly than facilities in general. Proximity is not causation - facilities cluster where population, airspace, and reporting infrastructure do.
Do sightings report what the patents predict?
The temporal and spatial layers ask when and where; this asks what. Patents describe capabilities - propulsion, sensors, directed energy, stealth - and each capability class predicts a specific physical signature. This crosswalk maps those predicted signatures onto the observed behaviour catalogued in DECUR’s documented cases, so you can see which cases actually report the signature a suppressed-capability class would produce. It is the semantic bridge that lets you ask the per-case question, deliberately matched on what is observed rather than where a patent was filed - because a secrecy order obscures location, not the physics. It is a curated mapping, not a statistical test; the rigor stays on the temporal spine.
How much capability is under active suppression, and when?
The final axis returns to time, but measures the suppression itself. Every patent in this dataset was placed under a 35 USC §181 secrecy order; this charts the suppression load- how many orders were simultaneously active each year - and surfaces the 38 that were never rescinded, the capability the state is still withholding. Because a secrecy order’s imposition and rescission dates survive the obfuscation that hides its assignee and location, this is the one dimension the disclosure paradox cannot erode - the most robust signal to lay against the sighting-density curve at the top of the page.
What survives every mundane-explanation pass?
The four axes above each explain a stratum of the record. This final view turns them on the cases themselves: it subtracts the explicable strata - cases near a known facility, and cases whose observed signature lines up with a capability under active secrecy suppression that year - to isolate the rigorously-unexplained residual. These are the high-anomaly cases far from every known facility, reporting a characterised signature, yet temporally decoupled from any matching patent activity: the falsifiable short-list. A coverage-confidencegrade then splits that list by regional facility density - separating genuine isolation inside a well-mapped field from cases whose empty surroundings are merely a hole in the map (orbital, open-ocean, remote). It is a subtractive lead-generation lens, with two hard caveats - the facility set is geographically incomplete, so “far from a known facility” is not strictly “isolated”, and “decoupled” says nothing about capability the state never patented at all.
Does the engine re-derive a signal we already know?
Every panel above surfaces a candidate signal with no answer key. This one is the opposite: a calibration check. The historical record robustly documents UAP activity clustering at nuclear facilities, so we run the same spatial proximity machinery that drives the residual filter against that known relationship. If it independently re-derives the nuclear signal, that earns trust in the machinery where there is no ground truth - and if it does not, that is a limit worth stating plainly. It is a sanity check on the method, never evidence about the phenomenon itself.
Does suppressed capability lead or lag the sightings?
Everything above measures whether two records co-occur. This is the discriminating test: which way the arrow points. Co-occurrence alone is weak - the Cold War aerospace boom inflates the patent record and the sighting record at the same time. So for each capability class we run the patent lifecycle as a chain - filing, then secrecy imposition, then sighting - and cross-correlate the detrended year-over-year series to ask whether suppressed development leads the sightings (built, then observed) or lags them (observed, then patented). Every peak is scored against a circular-shift null and FDR-corrected. The honest result is the point: applied with full rigor, the co-occurrence intuition does not survive - a discipline on the data, not a confirmation of it.
Do the named corridors and windows hold up?
The analyses above are open-ended searches. This one is targeted: it takes the specific corridors and time windows named in the source claims - Peenemunde 1943-46, the Bethpage–Bell Labs corridor, the Southern-California aerospace cluster, France 1954-56 - encodes each as a facility set, radius and year window, and asks one falsifiable question per preset: do sightings inside the corridor concentrate in the window beyond what the corridor’s own all-time sighting cadence predicts? Each is scored against a circular-shift permutation null. A preset that fails is the most useful output: a plain, falsifiable refutation of that claim. The test is year-resolution only - the static sightings binary drops month and day - so it is a year-window test, not a calendar same-day test.
What this can establish
A reproducible, falsifiable picture of where on the timelinesighting intensity and state-suppressed capability development rise and fall together - turning a vague “the patents and the sightings must be related” intuition into specific, testable periods.
What it cannot establish
It cannot prove causation, a terrestrial-technology origin, or an exotic origin. The same pattern is consistent with several hypotheses, and confounders are everywhere. Its honest value is as a lead-generation engine pointing rigorous investigation at the highest-signal periods in a large, noisy joint dataset.