We talked to 14 Go developers powering large-scale systems to ask how they actually handle observability. What surprised us: instrumentation often takes months, tracing is useful but hard to implement cleanly, and most teams are overloaded with telemetry data and struggling to keep costs down.