Methodology
How the least scientific test works 🔬
There is no language model, medical model, or hidden expert review. The site uses a small deterministic rules engine designed solely to create a funny, repeatable result.
Step 1
Normalize the input
The optional cat name and required behavior description are trimmed. The two strings are combined to form the input used for the result.
Step 2
Create a stable seed
A compact 32-bit string hash turns that input into a repeatable number. The same name and description produce the same seed and therefore the same result.
Step 3
Match a playful category
The lowercased behavior text is checked for keyword groups related to midnight zoomies, tiny-predator behavior, staring at walls, gravity experiments, landlord demands, orange-cat energy, and boxes. The category with the most matches wins; seeded selection breaks ties.
Step 4
Calculate the “crazy level”
The seed creates a base score from 35 to 82. Each matched high-energy word adds six points; each mild word subtracts five. A small seeded variation from minus three to plus three is applied, and the displayed result is limited to 12–99%.
Step 5
Choose the copy
The seed selects a verdict, diagnosis, personality type, explanation, and recommendation from handwritten templates associated with the matched category.
Safety override
Before showing a joke, a separate limited phrase check looks for text that may describe an urgent health problem. When it matches, the site suppresses the comedy result and recommends veterinary help. This check is intentionally cautious but is not comprehensive and cannot determine whether an animal is healthy.
How to interpret a result
The percentage, category, and “diagnosis” have no scientific or clinical meaning. They are deterministic comedy labels, not a probability, assessment, or veterinary recommendation.
Method verified against the application code: September 5, 2026.