Watch parties for a film streaming service

Your film streaming service is adding watch parties. Up to 8 friends in different homes watch the same film together, each on their own TV or laptop, and talk in a side chat. Anyone in the party can pause, play or jump to another point in the film.

At peak there are about 200,000 parties at once. Viewers notice when their screens drift more than about 1 second apart. Networks add 20 to 300 ms between a device and your servers. Now and then two people press pause or jump within the same second.

What does a device send when someone presses pause, play or jump? How does every screen end up at the same point?