render-the-candidateslisted

# Render the candidates Words like *left*, *longest edge* and *alongside* each name two things: an edge of the device or an edge of the display, the long side now or the long side once it turns. Two people can agree on every sentence and mean different screens, and neither can see which reading the other took. A third paragraph is judged against a sentence too, so it repeats the round with better vocabulary. ## Procedure 1. **Say the numbers first.** Whatever the disagreement is measured in (dp, pixels, aspect ratio, count), state the actual figures for each reading, in one line each. Often this alone ends it: "a side edge is 891dp upright and 411dp turned" is not a matter of opinion, and one reading usually stops being plausible the moment its number is written down. 2. **Enumerate the readings, and cap them at three.** Two readings are distinct if they differ in **anything under dispute**: a number from step 1, or where a named element sits: which edge, which order, which alignment. Merge only when every disputed quantity and every placement is identical. Two screens can agree on an edge and a dimension and still order or align their contents differently, and merging on the coarser test discards a candidate before anyone has seen it, which is the failure this skill exists to prevent. Name each in the user's own words, not yours. The cap is a budget, never a licence to drop the fourth quietly. If more than three survive the merge,