Entity interpolation: smooth motion by looking back
You know your own next input. You do not know what another player will do next.
Remote movement arrives in pieces
A server sends snapshots at discrete times. Drawing a remote player at the latest received position can make movement step from one update to the next. More snapshots can reduce the distance between steps, but updates still take time to arrive.
Render between known states
Interpolation keeps a small buffer and draws a position between two received states. Unlike extrapolation, it does not invent a future position. It intentionally displays a slightly older moment so that two known points are often available.
The trade-off is real
A buffer can hide uneven delivery, but it adds presentation delay. If snapshots arrive too late or are lost, interpolation may run out of useful points. A smooth player is not necessarily a current player.
Try it in the opponent view
Lower Server snapshots to 5 /s and replay a dodge. Toggle Interpolation and scrub slowly. Lag Arena’s opponent pane shows your player as a second client might see it. The target render time is behind nominal downlink delay by one snapshot interval plus the jitter allowance, and it shares the same simulated downlink conditions.
Separate model time from network time
This demo uses a shared simulation clock for the projectile. It does not implement clock synchronization, adaptive interpolation buffers or remote projectile replication. Use it to inspect a presentation trade-off, not to infer exact peeker’s advantage in a commercial game.
Further reading
The experiment and explanation above were written for Lag Arena. For a deeper technical reference, read Gabriel Gambetta’s multiplayer networking series.
Model review: October 31, 2026. This is an educational simulation; it does not reproduce a named commercial game.