THE UAP BASE-RATE PROBLEM.

There are two easy ways to be wrong about UAP statistics.

The first is to act as if every unexplained case begins at fifty-fifty: ordinary object or extraordinary technology, place your bets. The second is to point at a pile of resolved balloons and satellites and declare the remaining unresolved cases statistically dead on arrival.

Both approaches throw away information.

The boring baseline is enormous.

AARO's August 15, 2026 reporting trends list 941 closed-case resolution outcomes. Balloons account for 468. Satellites account for 345. Together they represent 813 cases, or approximately 86.4 percent of the closed cases in that table.

Add UAS, birds, aircraft, rockets, atmospherics, sensor artifacts, and other ordinary categories, and the broad lesson becomes difficult to ignore: the environment is full of things that can look unfamiliar when seen briefly, at distance, through imperfect sensors, under strange lighting, or without good geometry.

That is the base-rate problem. If someone reports a moving point of light at dusk, the existence of satellites is not a cynical debunking trick. It is prior information.

But “closed cases” are not a random sample.

Now for the part most internet arguments manage to misplace.

A case becomes closed because investigators believe they have enough evidence to resolve it. Easy cases are therefore more likely to enter the closed-case pool. A balloon with matching wind drift, a satellite whose predicted path lines up with a sighting, or a known aircraft with transponder history can be closed. A poorly documented observation may remain unresolved not because it is more exotic, but because the information needed to identify it no longer exists.

This means the closed-case proportions can be biased toward phenomena that are easier to confirm. The set of unresolved cases is conditioned on failure to resolve. That changes the statistics.

Bayes without the incense.

Bayesian reasoning is often invoked in UAP debate as if saying the word should end the conversation. The actual logic is less theatrical.

We begin with prior probabilities informed by how often candidate explanations occur. Then we ask how likely the specific evidence would be under each candidate explanation. A satellite prior may be high, but if a report includes reliable low-altitude range data, rapid maneuvering, and synchronized radar that excludes an orbital trajectory, the likelihood of the evidence under the satellite hypothesis collapses. The posterior probability should collapse with it.

Likewise, a rare hypothesis does not become likely merely because common hypotheses have not yet been proven. Failure to identify is not positive evidence for aliens, secret craft, interdimensional tourists, or whichever noun has acquired the most caffeine that week.

What the 86.4 percent figure really tells us.

It tells us that balloons and satellites are not fringe explanations. They are major generators of reports that reach an official UAP process and later get resolved.

That should change investigative order. When the geometry and timing are compatible, analysts should check satellite ephemerides and balloon possibilities early rather than after the extraordinary explanation has accumulated a fan club.

It also suggests a useful research question: which properties make a report survive ordinary identification?

Does unresolved status correlate with missing metadata, shorter observation time, single-sensor collection, unusual morphology, high apparent speed, or simply older records with poor provenance? That question is more informative than treating “unresolved” as a single physical category.

Unresolved is a workflow status.

This distinction deserves to be printed on every anomalous-phenomena database.

“Unresolved” may mean:

• the object is genuinely unusual;
• the data are internally contradictory;
• the relevant sensor data were not preserved;
• range is missing;
• timestamps are unreliable;
• no matching ordinary object could be found;
• multiple hypotheses remain viable;
• investigators have not completed the case.

Those are epistemic states, not species of aircraft.

The morphology numbers need the same discipline.

AARO's current trend table shows orbs/round/spheres and lights dominating reported morphology. Tempting conclusion: the sky is full of spherical craft.

Alternative conclusion: distant unresolved point sources tend to become circles and lights because optics, focus, glare, diffraction, display processing, and human categorization are cruel to fine geometry.

Neither explanation should be declared by morphology counts alone. The useful next analysis is to cross-tab morphology with sensor type, estimated range, day/night condition, angular size, focus state, and eventual resolution. If “orbs” disproportionately occur near the resolution limit of sensors, that tells us something. If they remain spherical at close calibrated range across multiple sensors, that tells us something else.

Build a survival curve for mystery.

One of the strongest datasets AARO could publish would show how cases move through investigation over time. At intake, how many reports fit each candidate category? After one week? After metadata retrieval? After satellite checks? After radar correlation? Which classes fall away first, and which persist?

That would let us distinguish hard to identify from hard to explain physically. Those are not the same problem.

A standard we can actually use.

For future Cyberdelia UAP analyses, every case should begin with a candidate-hypothesis table. Each hypothesis gets a prior based on occurrence and context. Each observation then updates the table. Missing evidence is listed as missing rather than silently converted into support for whichever explanation has the better soundtrack.

That is less exciting than belief warfare. It is also how a mystery eventually becomes knowledge.