Multiplayer Gaming

Overview

Cross-platform multiplayer infrastructure — matchmaking, lobbies, and server-authoritative netcode with client-side prediction — built once and reused across every multiplayer title we ship.

The Problem

Every new multiplayer title was rebuilding matchmaking, netcode and anti-cheat hooks from scratch, because those systems are genuinely hard and were treated as a per-game feature rather than shared infrastructure — which meant the second and third titles took nearly as long to ship as the first.

• FPS Games and an earlier strategy prototype had each solved client-side prediction independently, with subtly different bugs in each.
• Matchmaking logic was duplicated per title, so a fix in one game's matchmaker never reached the other.
• Anti-cheat hooks were the last thing added to each title, under launch pressure, which is exactly when they're most likely to be wrong.

How We Solved It

We generalised the netcode built for our own shooter and strategy titles into standalone infrastructure — skill-based matchmaking, dedicated server orchestration spun up per match, and server-authoritative simulation with client-side prediction and reconciliation, all cross-platform from the start.

• Skill-based matchmaking with configurable party and latency constraints.
• Dedicated server orchestration, spun up and torn down per match.
• Server-authoritative simulation with client-side prediction and reconciliation.
• Cross-platform lobbies — PC, console and mobile in the same match.

Key Decisions

• Extract from working code, not design from scratch. The FPS title's netcode was already proven under real players, so generalising it was lower-risk than a clean-sheet rebuild.
• Anti-cheat hooks live in the shared layer, not per-title, so a fix benefits every game that uses the infrastructure at once.
• Dedicated servers are ephemeral, created per match and torn down after, rather than a persistent pool that idles between matches.

The Result

The next multiplayer title we ship reuses this layer instead of rebuilding it, which is what's let build-to-launch time shrink with each new title rather than staying flat.

The first title built on the shared layer after FPS Games needed a fraction of the netcode-specific engineering time the original title did.