event-schedule-design

Standardize session lengths first, then place fixed points like keynotes, then breaks and lunch, then match sessions to rooms by expected audience size, and finally lay remaining talks hardest-constrained first. Collect every speaker's availability window before slotting anything.

Four distinct types: a session outside its room's availability window, a session conflicting with speaker unavailability, two sessions overlapping in one room, and a speaker double-booked across simultaneous sessions. For multi-speaker talks, the valid window is the intersection of all co-speakers' availabilities.

The default is a slot-internal buffer: publish slots longer than the talks they hold and brief speakers on both numbers. Whatever buffer you choose must clear the actual walking time between rooms; a zero-gap grid is ruled out when rooms are on different floors.

Release early and iterate is the default: treat the published grid as a living document revised in place. Lock-late publication is ruled out whenever attendees travel, since a grid published weeks before the event cannot inform flights booked months earlier.

The venue decides, not an efficiency ranking. Stagger starts when simultaneous room changes would jam a corridor, as PyCon US does across five rooms; synchronize when the room count is small enough that everyone moving at once still clears.

Do not use it to decide track count or session-format mix, which belongs to event-format-selection, nor for live day-of execution like responding to a no-show, which belongs to event-run-of-show. It only lays already-selected talks into the grid.