The Invisible Jungler

Analyse partially hidden game states and infer where the enemy jungler is likely to be, using evidence rather than guesswork.

Due: 19 April 2027 · Weight: 20%

The brief#

Analyse partially hidden game states and infer where the enemy jungler is likely to be, using evidence rather than guesswork.

You will be given several checkpoints, each built from a real (or realistic) partial game state: replay screenshots, minimap states, event timers and lane states, at a moment where the enemy jungler is not visible. None of these are trick questions with a single hidden correct answer waiting to be revealed. The point of the assessment is the reasoning, not the guess — which is why the marking below rewards a well-supported inference that turns out to be wrong over an unsupported one that happens to be right.

What you submit#

For each checkpoint, answer all five of the following, in order:

1. Where is the enemy jungler most likely to be? A specific location or small area, not “somewhere on the map.”
2. What evidence supports that inference? Cite what’s actually visible in the checkpoint — wave states (Week 3), last-seen location and time, missing camps, ward timers, tempo spent elsewhere (Week 6) — not a general impression.
3. What alternative locations remain plausible? At least one. If the evidence only supports one location with no live alternative, say so and explain why the evidence is that decisive — that’s rare, and it should be argued, not assumed.
4. What information would change the conclusion? Name a specific piece of evidence — a ward placed in a specific spot, a kill notification, a recall — that would move you from your answer to one of the alternatives, or confirm it.
5. Given the uncertainty, what should the player or team do? A concrete action (send vision, delay a commitment, group before contesting), not a restatement of the inferred location. This should read like Week 7’s map-state exercise: a decision made despite acknowledged uncertainty, not a decision that depends on the guess being correct.

What the criteria reward#

Evidence and citation rewards pointing at specific, checkable details in the checkpoint over a general impression of where a jungler “probably” is.

Quality of inference rewards reasoning that connects the cited evidence to the conclusion — the chain from wave state or timing to location — over an inference that simply asserts a location the evidence happens to fit.

Handling uncertainty rewards naming plausible alternatives and what would resolve between them, over presenting a single guess as though it were certain. This is where the assessment penalises confident guessing regardless of whether the guess turns out correct.

Use of course concepts rewards explanations built from wave states (Week 3), jungle tracking and staleness (Week 5), tempo and opportunity cost (Week 6), or the observed/inferred/missing/stale distinctions (Week 7) — over commentary that doesn’t name what evidence is actually doing the work.

Decision quality given available information rewards a recommended action that makes sense given what’s known and unknown at that checkpoint. A cautious action justified by real uncertainty scores higher than a bold one that only works if the guess happens to be right — and a correct location paired with a poorly reasoned decision does not earn full marks here.