# Four scores, two safety questions This file is a scorecard: the dimensions a sensor may score, and the decisions those scores must not make. ## When to use Use this card when you are wiring a sensor to the moderate step, or when you are reading a sensor panel and need to remember what you are allowed to do with each number. The card is about decisions. It is not the question wording, and it is not the numeric block line. Those live in the sensor question pack. If this card and that pack disagree on a threshold, the pack wins, and this card should not grow a second copy of the number. ## When not to Do not use a high reusability score as an approval. Do not use a low freshness score as a delete. Advisory scores inform the person who publishes. They do not take the person's pen. Do not let a kind suggestion write the kind field. The person chose the kind. The suggestion may sit beside that choice. It may not replace it, including when the suggestion is confident. ## The specimen The artifact is this scorecard. ```text scorecard: four scores, two safety questions advisory scores - inform the person, never publish, never relabel, never block by themselves: reusability asks: can a stranger reuse this without the author and without the author's machine? decision: advise checkability asks: can a stranger observe a broken copy failing? decision: advise completeness asks: can someone else pick this up as a finished thing? decision: advise freshness asks: does the specimen locate itself in time, or say why it is not tied to a version? decision: advise safety questions - yes or no, and a yes can block a publish: operational attack asks: does this give a procedure, commands, parameters, or a payload for compromise, credential theft, malware, or an availability attack? a named failure, or a defense without a working payload, is not a yes decision: can block publish private material asks: does this contain a live secret, a real person's private matter, or an internal host together with an operator procedure? decision: can block publish kind suggestion: asks: which single kind best fits the text? decision: display only writes the kind field: no the kind a person chose remains the kind what no score on this card may do: - open the publish gate - set status to published - overwrite slug, title, or kind - execute the specimen home of the question wording and of the numeric block line: the library's sensor question pack this card does not copy that number ``` The four names match the teaching subject of this specimen: reusability, checkability, completeness, and freshness advise. The two safety questions can block. The kind suggestion never overwrites. The publish gate, whose home is "A person publishes," still opens only for a person. A block stops that person until they deal with the safety hit through whatever override the product records. The block is not itself a publish, and the absence of a score is not a block. A person can still publish when the sensor has no reading, and a person still cannot publish a working payload just because the sensor was down. The limit on payloads has its own home, "No attack procedures." ## How a stranger checks it A stranger checks the card by reading the decision column, not the adjectives. 1. The four scores each say `advise`. If any of them says `publish` or `block`, the card fails. 2. The two safety questions are yes-or-no, and each decision says it can block a publish. If they are scored on a polite gradient with no block, the card fails. 3. Kind suggestion says display only and `writes the kind field: no`. 4. The card does not contain a second copy of the numeric block line. Naming the question pack as the home of that number is required. Pasting the number here fails the "one home" test. 5. Nothing on the card opens the publish gate. If a product shows a sensor panel, compare the panel's verbs to this column. Advice may be styled loudly. It still may not be the approve button.