Spatial layout has no automatic typesetting; overlap is constrained only by rules

What it is

When desktop windows multiply, the system snaps to edges, avoids a total bury, and offers a taskbar. In space, two panels’ centres are set by grab or follow; nobody runs a flow layout on them. If they overlap, they overlap — unless the product writes a rule. Spatial overlap rules are not taste. In an environment with no typesetting engine they are the only thing that stops two readable surfaces fighting for the same line of sight. Space without rules looks free and reads like a table of uncollected paper.

Why it happens

A 2D layout engine handles rectangles in a rectangle: constraint solving, anchors, flow wrap. Overlap is treated as a bug or digested by an explicit z-order. In 3D a pose has six degrees of freedom, solving is expensive, and “yield in which direction” has no unique answer — up may leave the view, back may enter a wall, sideways may leave reach. So the engine is absent and overlap becomes the default runtime accident: lazy follow drives two carry-on panels into one eccentricity, a world-locked notice posts onto a menu already pinned, a grab release lands directly in front of another plate. Without rules, people solve with the body: a side step, a hand to push aside, paying the layout bill in posture. Rules can be crude: a minimum angular gap, mutual exclusion at one depth, later-opened yields to earlier — crude, but stronger than none.

Studying it

Have people open several panels at once for a comparison task, under no rule, a minimum-angle push-apart, and a ban on same-line-of-sight overlap. Log overlap events and spontaneous tidying.

Independent variables: rule kind, panel count, whether locking modes are mixed. Dependent variables: overlapped area along the line of sight, fraction of readable type that is covered, count of spontaneous tidying, completion time.

This is collision detection for information surfaces, not rigid-body physics — the question is whether readable surfaces are collinear, not whether bounding boxes intersect. Two boards geometrically separate but fully overlapped on the retina count as overlap for reading.

Where it stops holding

A short solo task with one panel almost never hits overlap, and rules look superfluous. Games that plaster a wall with posters are scenery, not readable UI. When manual placement is the task (hanging a show), rules get in the way and should be switchable off. Automatic push-apart with no transition makes panels look alive and fights object persistence. In multi-user sharing, one person’s push-apart is another’s sudden displacement; the rule must be shared, or must only push private layers.

Applying it

• Write a minimum angular gap for readable panels: below it, push apart or place side by side. Do not silently overlap.
• When locks are mixed, a carry-on layer must not park directly in front of a world layer; follow includes avoidance, not only body tracking.
• Push-apart needs a visible displacement and an undo, so it does not read as “the board ran away”.
• How to check: open the maximum number of panels the product allows and look along the main line of sight. If two bodies of readable type share one visual angle and no rule split them, the missing typesetting engine has not been replaced by a rule.

Related

• Same group: N3.14.1 Space got larger; the readable angular range did not · N3.14.2 Stacking panels in depth does not increase effective capacity · N3.14.4 The extra placement freedom is also a tidying cost
• Nearby: N3.07 Content at the Edge of the Field of View · N3.08 World-locked, Body-locked, and Head-locked
• Search terms: spatial overlap · layout constraints · UI collision