Method
The rules of the game
Any UAP evidence collection needs an answer to three questions: what counts, how it's weighted, and when it changes. Here are all three, so anyone can check the analysis instead of having to believe it.
Chapter 1
Why Tehran isn't equivalent to Communion
Three evidence tiers (S, A and B): a case with sensors and military witnesses does not weigh the same as a single-witness one, and the tier does not say what the case was
Not all cases are worth the same as evidence. Mixing tiers muddies the analysis and makes it look as if there were more evidence than there is, and the public and the press mix them up often.
| Tier | Category | Examples |
|---|---|---|
| S | Strong evidence: a contemporary instrumental record (radar, sensor, film or photo of known provenance) that corroborates several independent witnesses | USS Nimitz Tic-Tac, Belgian Wave (F-16), Lake Huron F-16 shootdown, JAL 1628 |
| A | Institutional evidence: several identified independent witnesses, or contemporary official documentation, without an instrumental record to corroborate them | Hessdalen lights, Hudson Valley Boomerang, Trans-en-Provence, Falcon Lake |
| B | Limited evidence: a single witness, testimony gathered years later, or an account without contemporary primary documentation | Communion (Whitley Strieber), Maury Island, Bonnybridge, Gulf Breeze |
Three rules set the boundary between tiers. Without a contemporary instrumental record, a case goes no higher than A, however many witnesses it has. Testimony gathered years later does not count as contemporary. And the tier measures the strength of the evidence, not what it was: an S case can have an ordinary cause.
Documents have no entry or tier of their own: each one appears on the case or actor profile it supports, with a paragraph on what it is, who issued it and what it proves.
A Tier B "evidence" should not be used to support conclusions that require Tier S. The scale (descended from Hynek's close-encounter categories) measures only the strength of the evidence, not what the case was.
Three different measures
Each case is described with three distinct measures, spread across the site's views. They are easy to conflate because all three sound like «quality»:
- Tier (S/A/B): the strength of the evidence. Shown as a badge on each case and as the marker color on the map.
- Probability (0–100%): how much of the case remains unexplained: the share of the split of explanations that no known cause covers («indeterminate» plus the two non-human explanations). It is not written by hand: it is computed from that split, so it never contradicts it. Shown as «%» on each case and as the marker size on the map.
- Split of explanations: what it was: how the case splits across six explanations that exclude one another. see /probabilidades →
They are independent: a well-documented case (Tier S or A) can have a possible hoax as its most likely explanation, and a Tier B is not a hoax for that reason.
Chapter 2
Why case number 50 adds nothing new
A case with sensors can change which explanation weighs most; case 50 of the same kind only reinforces the one already dominant
Each case adds its full split, with no hidden discounts. What changes is how much that split tells us: an institutional case with a sensor gives a sharp split that can move weight to an almost empty explanation; the fiftieth case of the same pattern only reinforces the explanation that already dominated.
Shifts the split between explanations
- ·Tier S or A with several sensors (Tehran, Nimitz)
- ·New class of evidence (Hessdalen, Lake Huron)
- ·Case that contradicts the established pattern
- ·Official sensors and matching video
Only reinforces the dominant
- ·An isolated Tier B contactee (Communion)
- ·Case 50 of the same kind
- ·A contactee prediction that did not come true
Isolated contactee cases split mostly between ordinary causes and indeterminate, so they do not change the archive's split. Institutional cases with sensors (Tier S or A) can.
Chapter 3
Why probabilities sum to 100%
In each case, the six explanations exclude one another and share 100%, so they can be compared
In each case, the explanations exclude one another: the case had a single real cause, and the uncertainty is split among the candidates until it sums to 100%. Added across all cases, they can be compared: you can say which explanation accounts for more cases. On the site, each case is shown under a single explanation. Ordinary ones are named by what the object was (a star or planet, an aircraft or balloon, a rocket or satellite…), a natural phenomenon or a possible hoax; the two non-human ones are grouped, and the share a case cannot assign is spread across the explanations it does support, so no case is left unclassified.
Each case's probabilities are structured judgments, not measured frequencies: being comparable does not make them true. They say which explanation fits best with what is known about each case, not which one is correct. There are two ways to add up the cases. One adds the probabilities of each explanation: it gives the expected number of cases per explanation and spreads the indeterminate share in fractions. The other counts each case once, under its most likely explanation, with 'Indeterminate' as its own category: that is what the home page, /probabilidades and the /cases filter use. The two give slightly different figures because they measure different things, not because they contradict each other.
On /probabilidades, ordinary explanations open into the specific object, and 'pending review' marks the cases not yet checked against any source: their class is the inherited one, not a conclusion. On each case page, the Probability is the share of the split no known cause explains (the indeterminate part plus the two non-human explanations), and the tier measures the strength of the evidence, not what it was: a well-documented case can have a possible hoax as its most likely explanation.
The detail of each explanation, with its figures and most representative cases, is at /probabilidades →
Chapter 4
What is settled vs the frontier
Most cases have an ordinary cause (mostly aircraft, balloons, celestial bodies and satellites); the non-human explanations are a minority and concentrate in the best-documented cases.
The model was rebuilt so the numbers can be compared. Before, each hypothesis was a separate claim ('at least one case is X'), and its percentages neither summed to 100 nor competed: you could not say which explanation was more likely. Now each case splits 100% across the same six explanations, which exclude one another, and the archive total is split the same way. Each explanation combines what the object was and what the state did, so 'non-human and hidden by a state' is its own class. The previous model's hypotheses remain inside each explanation, and 'non-human entities' (the sum of the two non-human ones) and 'heterogeneity' are calculated from them.
The full split, the six explanations and the most representative cases of each are at /probabilidades →.
Chapter 5
What would move the analysis now
Lake Huron isotopic analysis, verifiable recovered material, formal acknowledgment by a new country
- +++Independent isotopic analysis of physical residues published
- +++Recovered material with verifiable photos
- +++Published analysis of the Lake Huron fragments
- +A new country formally acknowledging the phenomenon
- +Another case like Tehran, with sensor data
- ~0A new contactee with a detailed cosmology
How it adds up
How each new case changes the figures
Each case splits 100% across the six explanations. That split is a declared judgment, not the output of a formula: the case analysis decides how well each explanation fits the evidence, and the split stays visible on the case page, under 'Distribution of explanations'. They are reasoned judgments, not measured frequencies: they say which explanation fits each case best, not which one is correct.
The archive total is computed automatically: it adds up the splits of every case and rounds them so the displayed percentages sum to exactly 100. Because it is a sum, it holds even when cases resemble one another. Each new case enters with its split and the total is recomputed when the site is published, with no manual adjustment. In the charts, the share a case cannot assign is spread across the explanations that case does support, so no case is left unclassified.
Now see the result: how the archive splits across six explanations
The full methodology is on this page, with no external documents.