Machine-based anti-cheat has one reliable blind spot, and Counter-Strike found it in 2020: the most consequential form of cheating that year involved no injected code, no spoofed input, and no player at all. It involved coaches. The scandal is worth revisiting on a site about VACnet and automated judgment, because it is the clearest example of a case the machines would never have caught — and of what human institutions are for.
The bug, in one paragraph
CS:GO’s spectator system had a long-standing quirk: a coach or spectator viewing a match feed could see information slightly beyond what was fair — a glimpse, a delay quirk, a free look around the map while a round was live. The trick had been known for years and treated as etiquette rather than law. Some coaches used it sparingly. Some used it industriously, calling rotations with a tactical picture their players could not legally have. Nobody had ever been punished for it, because nobody had seriously proposed that watching was a weapon.
September 2020: thirty-seven at once
The reckoning arrived fast. HLTV reported that on 28 September 2020 the Esports Integrity Commission banned 37 coaches for abuse of the spectator bug, with sanctions ranging from roughly four months to three years. The scale was itself part of the shock: this was not one cheater caught in the act, but an entire professional habit retrospectively reclassified. Reports at the time noted ESIC had reviewed only a slice of the available demos, which meant the number of sanctioned coaches was a floor determined by investigation capacity, not by guilt.
And the loudest argument was about the exact thing every anti-cheat system must eventually face: when did the rule change? Coaches who received the longest bans had used the feed most aggressively during its ambiguous period. Their defense — we were winning with a feature, not a cheat — was not unreasonable. The commission’s answer, in effect, was that an exploited ambiguity is still an exploitation, and the platform rules belong to the platform’s watchdog.
Why VAC would have seen nothing
Run the scandal through the automated stack and it dissolves. VAC scans processes; a coach running the game as a spectator has an honest process list. VACnet models player behavior; the suspicious party was not on the server at all, and the behavior in question was communication, not aim. VAC Live ends matches; it would have had nothing to end. The only signal anywhere in the system was that an unusual number of very well-informed rotations kept happening, which is not the kind of datum a signature scanner reports.
That is precisely the space institutions like ESIC occupy alongside the software: they police the layers the telemetry cannot see — who was allowed to know what, and who was paid to know it. CS2 inherits the same boundary. Its sub-tick data and labeling portal will get ever better at judging mechanics, and will remain structurally unable to notice a manager with a monitor.
The long tail
The coaching saga did not end with the bans. ESIC later ran a historical investigation covering periods before the rule was formalized, publishing updated charge outcomes as late as 2022, because retrospective integrity work always discovers older cases. Meanwhile the practical remedy arrived from an unexpected direction: the general move of elite CS to platforms with dedicated coach protocols and LAN rulesets, where the spectator question is settled by contract rather than by geometry.
For anyone tracking CS2’s anti-cheat era from a machine’s point of view, keep the coaching bug on the shelf as the cautionary exhibit. The best detector in the world measures what it was pointed at. Everything else — every free look, every well-informed call, every $4.5 million bet — is left to the humans.

Leave a Reply