Section 9: The Gauntlet - Interviewer Deep-Dive Challenges

Executive Summary: The interviewer goes after the decisions you sounded surest about: peer-to-peer play, a million-viewer final, two tabs sending the same ply, an actor-free Redis design, racing rating updates, and a disputed flag. Each runs the same way: the probe, what it really tests, the answer with numbers, and the senior line.

The interviewer caps the marker. "Good. Now let me poke at it." The answer that works each time: name the decision, give the number behind it, and say what would make you change it.

Challenge 1: "Why Not WebRTC Between the Two Players?"

"Priya's moves cross two gateways and a game server to reach Omar. Connect them directly."

What they're really testing: whether you know the server is the referee, not a relay.

You: "With peer-to-peer, nobody neutral checks legality, owns the clocks or declares the result. Each client trusts the other's clock, so one edited line of JavaScript forges think time. Spectators can't watch a link between two phones, and strict NATs need a relay anyway. And the gain is small: in-region relay is p99 < 50 ms, gateway timekeeping already keeps transit off both clocks, and the whole platform pushes about 186 Mbit/s."

The senior line: "P2P removes a hop and the referee with it. Rated chess is the referee, so I'd only use P2P for a casual offline or LAN mode."

Quick check before moving on: