weight

QUESTION WORTH ASKING: how much a curious engineer, reading only the plain-words question of this inquiry, wants to know the answer — a question a real person would type, about something that matters or delights, specific enough to have an answer and open enough that the answer is not already known. Judge the question itself, as confirmed by what the rows show it was really about; a grand question whose statement answers a smaller one counts as the smaller one.

SURPRISE: how much the measured rows of this inquiry would make a well-read engineer stop — a result they would not have predicted, a name or number they would repeat to someone, a pattern that contradicts the easy guess. Judge the rows on the card, not the promise of the question or the adjectives of the approach; an expected result, however well made, scores low here.

CORPUS REACH: how much this inquiry answers something only this corpus can answer — it leans on what is rare about the archive (full history, joinable graphs and threads, several venues under one SQL surface, native citation and follow edges, embeddings beside timestamps) rather than on anything a site's own search box, a public API call or a well-known dashboard already gives. The more the result depends on holding the whole corpus as tables, the higher.

THE APPROACH TEACHES: how far this inquiry's approach carries a reader who could not have written the statement from the question to being able to write it — it names the relation and the columns that make the move possible, the one trick that does the work, and the traps (denominators, duplicates, what is counted and what is not). Count only teaching that is true of the statement on the card; length, praise of the inquiry, and claims the statement does not carry add nothing.