# Blame the Thing

> A light design check for adults, at https://blame-the-thing.skillsafe.ai/. Describe one everyday thing that tripped you up (a glass door you pulled that needed a push, hob knobs in a row, a self-checkout, a kettle) - what it is, what you were trying to do, and what you did and what happened. Five design checks say whether it was you or the design, and an embossed label to stick on the thing makes the next try go right.

URL: https://blame-the-thing.skillsafe.ai/
API tutorial: https://blame-the-thing.skillsafe.ai/api.html

## What it does

- Inputs: the thing, what you were trying to do, what you did and what happened (or a dropped .txt file), how often it trips you up (first time, now and then, every time) and whether others get it wrong too.
- The five checks, in order: signs (could you tell what to do and where?), mapping (do the controls line up with what they change?), blocks (were wrong moves blocked or made hard?), feedback (could you tell what happened, straight away?), undo (could you get back from a wrong move?). A check fails only when the description shows the problem; a check the description does not mention counts as holding up.
- Free, in the browser: keyword cues per check (for example "no sign", "in a row", "nothing happened") with a provisional score, a drawing of a handheld embossing label maker whose wheel carries the five checks with the failed ones in red, your own label in red, blue, black or green tape as SVG or PNG, the picture as SVG or PNG, Markdown and JSON downloads, and saved diagnoses with how the same thing scored before.
- Metered (one run on gpt-terra): each check held up or failed with a note, the score (two points per check that held up, out of 10), the verdict (first match wins: 4 or less = blame the thing; 8 or less = a bit of both; otherwise just a slip), the culprit (the first failed check), slip or mistake, an embossed label of at most 24 capitals and where to stick it, one note for the person and one for whoever made the thing. The page adds up the score again from the reply's own checks.

## Contract

POST https://api.skillsafe.ai/v1/app-api/run with string fields: task = "diagnose", thing, goal, happened, often ("first time", "now and then", "every time" or "not said"), others ("just me", "others too" or "not sure"), cues (the page's keyword cue lines, or "none"). The reply is one JSON object: {task, headline, rows: [{id, ok, note}] for discover, mapping, constraints, feedback and recover, score, verdict, culprit, error_type, error_why, label, label_where, for_you, for_maker}.

## Notes

Labels and notes only, not professional advice: never open, rewire, force or cover anything, and never stick a label over a warning, vent, display or control; in a shared, rented, work or public place, ask whoever runs it first. Built on the SkillSafe skill @wondelai/design-everyday-things (https://skillsafe.ai/skill/@wondelai/design-everyday-things, MIT): signifiers, mappings, constraints, feedback, conceptual models, the two gulfs, slips and mistakes, and a five-row quick diagnostic scored out of ten.