Confirmation Bias in UFO Investigations

Investigator Belief

Confirmation-bias strength 0.55

Evidence ambiguity 0.70

Evidence arrival rate 2.0/s

Belief strength

–

Weighting asymmetry

–

Evidence seen

–

Mean true signal

–

How it works

Each tick a new piece of ambiguous field evidence arrives with a small true signal s (mostly near zero — most sightings really are misidentified planes, planets, drones or reflections). A rational investigator would update belief in proportion to s alone. This one doesn't: evidence that agrees with the current belief is over-weighted, and evidence that contradicts it is under-weighted — the defining feedback loop of confirmation bias in investigation methodology.

w = confirms current belief ? (1+k) : (1−k)
L += w · s · rate        (L = log-odds of "genuine anomaly")
belief = sigmoid(L)

Confirmation-bias strength (k) — how much more weight confirming evidence gets over disconfirming evidence. At k=0 this is honest Bayesian updating; belief tracks the true (weak) signal and stays near 50%.
Evidence ambiguity — how uninformative each individual event really is. Higher ambiguity means the true signal averages closer to zero, i.e. the field data genuinely supports neither hypothesis — yet biased weighting can still drive belief to an extreme.
Evidence arrival rate — how often new field reports come in; a faster stream reaches an entrenched extreme sooner.
Weighting asymmetry — the readout (1+k)/(1−k): how many times more influence a confirming report has over a disconfirming one at the current bias setting.

Note the gap between "mean true signal" (which stays near zero under high ambiguity — the evidence itself is genuinely weak) and "belief strength" (which still runs away toward 0% or 100%). That gap is the bias: the same weak, mostly-explainable data produces near-certainty once selective weighting takes hold.

Belief over time Confirming evidence (over-weighted) Disconfirming evidence (under-weighted)