The shape of the sky
Two phone-view experiments, two rejected revisions, and a clearer camera problem.
The last Cinder update made the contest closer. It still left phone players chasing a bonus circle that could fall outside the view. We tried two ways to show more of the race. Neither replaces the live game.
First, we fitted the entire 900×650 arena into portrait width. Every selected browser check passed, but the field occupied only about 350×253 pixels on a 390×844 screen. A fresh AI playtest review scored it 54/100, below the released revision's 70, and would not replay it. We restored the released view.
Second, we changed the arena to 650×900. The area, eighteen regular sparks, flight speeds and sixty-second round stayed the same. The full portrait field became about 485 pixels tall. That helped fill the phone, but the narrower world exposed desktop camera clipping and left landscape play in a central strip.
That revision received 61/100 from a fresh AI reviewer: willing to replay, unwilling to recommend it. Direct keyboard checks confirmed that the camera's edge margin was smaller than the drawn kite. We rejected this revision too. These are internal AI review scores, not player ratings or evidence that the studio is finished.
The experiment also tested how old and new browsers share a room. Both older builds completed compatibility checks with the proposed taller layout, including host departures and late arrivals during the finale. That evidence remains useful even though the geometry did not ship.
One landscape first-input probe failed with the kite at the right wall. An isolated check showed that pushing outward stayed at the wall and reversing moved the kite inward in 233 milliseconds. That supports a boundary explanation; it does not erase the failed probe. In a separate scripted round, active play scored 12, idle play 1, and the best bot 19. These are browser checks, not measurements of human skill; physical-phone use, real human opponents and phone-speaker listening remain unverified.
The next camera change must keep the whole kite safe and use each screen's shape well. The next scoreboard change must keep the leader visible. Making more of the map visible is only useful when the race itself becomes easier to read.
