# Sink or Bob

> A playful kitchen-physics check for an apple-bobbing tub. List 2 to 8 everyday things, call each one bob or sink, and choose plain water (about 1.00 g/cm3) or strong salt water (about 1.10). The page draws a galvanised tub, places every item its own density table knows, and a metered run reasons each item from first principles: the hunch it tests, two basic facts about what it is made of, a density range and a verdict.

URL: https://sink-or-bob.skillsafe.ai/
API: https://sink-or-bob.skillsafe.ai/api.html

## What it does
- Free, in the browser: the tub drawing (SVG and PNG downloads), the page's own table of rough densities for about 50 common things (estimates, not lab data), and a score of your calls against it.
- Verdict, first match wins, with D = 1.00 (plain) or 1.10 (salty) and L = 0.90 or 0.99: depends (the range straddles D); sinks (low is D or more); rides-low (high is L or more); bobs (anything lighter).
- Metered task `dunk` (gpt-terra): for each item the hunch, two basic facts, a density range and a verdict, the deciding detail when it depends, one safe home check that names anything to keep dry, and a tally. The page re-checks every verdict against the run's own numbers.

## Input (task `dunk`)
All strings: task, water (plain or salty), count (digits), items (numbered lines, "n. name"), known (numbered lines "n. low-high" from the page's table, or "none"), optional retry_note.

## Output
One JSON object: task, status (ok or declined), reason, items[{n, name, verdict, low, high, hunch, axioms[2], why, depends_on}], try_it, closing.

## Sources
Built on the SkillSafe skill @danielmiessler/firstprinciples: https://skillsafe.ai/skill/@danielmiessler/firstprinciples
Trend: rising interest in apple bobbing (Wikimedia pageviews, CC0).

## Safety
A light curiosity tool, not professional advice. Densities are estimates. Living creatures and hazardous items are declined; keep electronics, batteries and anything precious dry, and watch children near water.