Why a stream lags behind the stadium, and where each second of the delay goes

Delay is deliberate, not accidental

It is tempting to assume that a stream lags because something is slow, when in fact most of the gap consists of buffers that were added on purpose. A buffer is simply a store of video held ahead of what the viewer is watching, so that a brief interruption can be absorbed without the picture stopping. Every stage in the chain keeps one, because every stage has to protect itself against variability in whatever feeds it.

Removing those stores would produce a stream that reacts instantly to the event and also to every network hiccup between the venue and the viewer. The delay is therefore the price of a picture that keeps playing, and the two cannot be optimised independently.

Segmented delivery is the biggest single contributor

Streaming systems generally cut video into short segments and deliver them as ordinary files, which is what allows them to use the same infrastructure as the rest of the web. A player must receive a whole segment before it can decode and play it, and it usually holds two or three ahead to guard against a slow download. That structure alone accounts for a large share of the total delay, and the delay scales directly with how long each segment is.

Shorter segments reduce latency but increase the number of requests, which raises overhead and makes the stream more sensitive to a slow connection. Choosing a segment length is therefore a straightforward trade between how current the stream is and how robust it is.

The other contributors add up

Encoding at the source needs to see several frames before it can compress efficiently, which introduces a delay before anything is sent at all. Packaging the encoded video into the delivery format adds another increment, as does each re-encode performed for a different device or quality level. Distribution across a content network adds transit time and, for the first viewer of a segment in a given region, the time taken to fetch it from the origin.

Finally the player has to decode and render, holding its own small store so that playback is smooth rather than stuttering. None of these is large by itself, but the sum is what puts a stream behind a traditional broadcast that took a shorter and more specialised path.

Low-latency modes and what they cost

Delivery methods designed for lower latency send parts of a segment as they are produced rather than waiting for the segment to be complete. This can cut the gap substantially, and it is why some services now sit much closer to real time than they did a few years ago. The cost is that the player has less video in hand, so a momentary drop in connection speed becomes a visible stall rather than an invisible one.

Services therefore offer this as an option rather than a default, since most viewers prefer a stable picture to a marginally fresher one. A viewer choosing a low-latency mode is explicitly accepting more interruptions in exchange for being closer to the action.

Why perfect synchronisation is unattainable

Different viewers take different paths through the distribution network, use different devices and sit behind connections of varying quality. Each of those differences changes the size of the buffers involved, so two people watching the same stream can be meaningfully apart. Any attempt to synchronise them must slow the fastest to match the slowest, which means everybody watches further behind the event than necessary.

Some services do exactly this for specific features that depend on everyone seeing the same thing simultaneously. For general viewing the trade is not worth it, which is why the notification on a phone will keep arriving before the moment on the screen.