lag mitigation Lag mitigation with constant speed simulations

[dhakan]

Hi! I'm in the process of developing an online multiplayer snake (4-8 simultaneous players in a grid based arena) with the stack involving Phaser in the client, Node in the backend with the networking protocol being websockets through socket-io. The progress made so far has been a prototype consisting of a dumb client and an authoritative server. The server proceeds game logic by 100ms and emits the result to the client at the end of each iteration of this game loop. The client is able to tell the server to change its direction to up, down, left and right, respectively, and the server will use this in order to calculate the next x and y position for this client (which is on the next tick). If the server notices clients colliding with walls, other clients or fruits, the involved client(s) is affected accordingly.

The issue with this "dumb client" approach is that the responsiveness for the client seems to be lost in not actually simulating anything locally.

From what I understand, to mitigate this lag for the client, client prediction can be applied. Since the movement is actually not caused as a direct action from a client but is always constant (you only influence direction in snake), this would mean that this 100ms game loop would have to be simulated in the client as well, with directional client input being respected and simulated directly in the client at the same time as it is being sent to the server for "verification". 

For some reason I cannot seem to wrap my head around this topic of the game having "constant speed". The examples I've read are mostly based on games where the movement is based directly on a fixed action, say "move me 10 px to the right", and not constant in the way snake is.

Does this “constant speed” design change how I should tackle lag mitigation techniques such as client prediction? What about entity interpolation/extrapolation? Any help or guidance would be greatly appreciated.