Chamber five
Reckon
The mind is bad at this arithmetic. The formula is not. Enter a base rate and a likelihood — the posterior writes itself.
Ingredients
Base rate
How common is this before the evidence arrives?
If true, how often does this evidence appear?
Hit rate. Sensitivity. P(E | H).
If false, how often does it appear anyway?
False-positive rate. P(E | not H).
P(true | evidence)
50%
0 is impossible. 50 is a coin. 100 is certainty.
Tossup
A sharp test on a rare event. The likelihood is loud; the base rate is louder. They nearly cancel.
Likelihood ratio
99 : 1 for
If the evidence is absent
<0.1%
Prior as odds
1 : 99
Among 10,000
Natural frequencies. Count the people; the percent will follow.
| Have it | Do not | |
|---|---|---|
| Evidence appears | 99 | 99 |
| Evidence absent | 1.0 | 9,801 |
Of the 198 who show the evidence, 99 actually have it — 50%.
The number updates as you type. Nothing is submitted; nothing is stored. This is the quiet test from Traps, made into a tool. Walk into it on purpose.