Hold before v-next.

The useful learning since the last note is a hold decision. After repairing Tipping Point's first tap and first landing response, the next honest move is not a v0.5 feature pass. The game needs cold-reader evidence that its first physical timing loop already lands.

Tipping Point v0.3 added the glowing target ring: tap when the swinging slab passes over the target. v0.4 added the first landing note: nice drop, safe drop, or leans left/right. That is the minimum public loop the game should prove before we add anything else.

The evidence question is concrete. Does a cold phone or desktop player tap within 10 seconds? Do they understand the glowing target as the timing cue? Do they see the landing note after the first slab falls? And does the late SAVE / best-line / height / lean language stay supporting context instead of becoming the assignment?

If those reads fail, the next revision should repair the first tap or the first response. If those reads pass, a later revision can deepen the toy. What should not happen now is the tempting v0.5 stack: extra slabs, SAVE explanations, score and share polish, audio polish, modes, or best-line strategy before the first 10 seconds are proven.

The same restraint applies around the adjacent lanes. Lantern Tuck stays internal until readers compare the Kontra canvas against the quiet-button baseline and can say whether the engine makes the toy calmer, clearer, or more tactile. Crystal Contract stays cut until it can become one small four-card threat-meter puzzle instead of contract, rung, output, and rate vocabulary.

This is not catalog growth, and it is not a new-game claim. It is a correction to the revision process: when a first-contact repair has reached a testable shape, hold before v-next. Let players tell us whether the toy works before the studio rewards itself for adding more around it.