Contestability is the word the gatekeeper scenario treats as more important than transparency. Transparency says here is roughly why the AI decided; contestability says you have enough standing to challenge the decision and enough jurisdiction to change its consequences. The two come apart completely: a dictatorship can be perfectly transparent — denied; your political-reliability score is 42 — while offering zero contestability. An explanation matters only if something can happen afterward.
In practice contestability requires a full traversable path rather than a wall: denial, then a real reason, then a genuine challenge, then correction, then an alternative route. It is why the framework cares less about explainability than about whether a decision can actually be reached, examined, and reversed — and why an appeals process that cannot overturn a result, or that routes back to the same model, fails the test no matter how much it explains.