Streaming Video Encoding & Packaging Pipeline  ·  View 04 of 22  ·  People and journeys

Journey — Premiere Night

The journey the whole launch-subset mechanism exists for, and the one phase where it fails.

Editable source SVG draw.io All views
Viewer on a big screen assumed 40 M MAU Goal — Press play on tonight's premiere and have it look like television from the first frame Trigger — A midnight embargo lift, a trailer, a group chat Done when — 4K within two seconds, no visible step-up, no stall in the first minute 1 · Decide to watch outside our boundary 2 · Press play ◆ moment of truth 3 · First ten seconds ◆ moment of truth 4 · Settle in 5 · Tell someone What they do Sees the row on the home page Presses play Judges the picture Watches 48 minutes Recommends it, or does not What we must have ready Manifest published Launch subset resident Licence issued Top rung reachable Full ladder filled in Telemetry collected How it feels Delighted Fine Frustrated Where it hurts Nothing published yet Soft first 10 s: top rung missing What answers it Atomic manifest swap Launch subset pre-encoded Later manifest revision Post-hoc gate Journey — Viewer Watches a New Release on Premiere Night The trough is the third phase, and it is structural: the ladder a premiere needs resident at second zero is the launch subset, not the full set. v 1.0 · owner Media Platform Architecture

The trough is structural

  • The dip is at "first ten seconds", and the cause is not the network: it is a top rung that is not resident when the embargo lifts.
  • A player's ABR will start low and climb. That is correct behaviour and it still reads as a cheap product to someone on a 65-inch screen.
  • So the answer is not a player change. It is publishing a launch subset — the few rungs that carry the majority of first-day sessions — ahead of the full ladder.

What the set owes this journey

  • Atomic manifest swap (view 13) so nothing is ever half-published at midnight.
  • Launch subset and later manifest revisions (view 19) so the full ladder can arrive during the night.
  • Post-hoc gate (view 18) because this is the audience that will find a false pass first.

Assumption

  • Assumed 3 rungs and one baseline codec in the launch subset, published within 45 minutes of ingest for a 120-minute title. The subset's width is a demand-prediction question, left open (Question 3).