Portal Mode: Designing Routes Through a Teleport

Portal Mode places two linked cells on the 16x16 board, A and B. A robot that slides across A immediately emerges from B and continues in the same direction under the same stopping rules. It does not turn, it does not pause, and it does not lose momentum.

Three details decide most rounds. First, the teleport fires once per move, so a robot that exits B and later crosses A again on the same slide will not teleport twice. Second, if a robot is already resting on the exit portal, teleporting is disabled and the sliding robot passes over A as if it were ordinary floor. Third, nothing else changes: robots still stop only at walls, board edges, or other robots.

Read those three rules as one sentence. A portal is a conditional shortcut between two lanes, and the condition is whether the far end is occupied.

The most valuable thing a portal gives you is reachability. In standard play a target is approachable only from the four lanes that pass through it, and if none of those lanes contains a stopper, the target is expensive to reach. A portal can add an approach that never touches the target row or column at all.

Concretely, suppose portal B sits two cells to the left of a wall on the target row. Any robot that enters portal A from any direction now has a path that ends against that wall. A column that was worthless a moment ago becomes the fastest approach on the board.

This is why scanning a portal board should not start at the target. Start at portal B and ask what lies downstream of it in each of the four directions. That answer tells you what the portal is worth on this particular board, and it is often worth nothing, which is equally useful to know early.

The occupied-exit rule is the mode's most underused tool. Parking any robot on portal B disables teleporting from A for every later slide, which means you get to decide whether the shortcut exists.

That matters when a portal is in your way rather than in your favour. If the natural route for your target robot crosses A and would fling it to the far side of the board, a single setup move parking a helper on B converts A back into ordinary floor and lets the slide finish where you intended.

It also does double duty. A robot parked on B is a wall like any other, so the same move that disables the teleport can supply the exact stopper your final approach was missing. When a portal round has a clean two-move solution, this is usually why.

Portal boards punish improvisation more than standard boards, because one unnoticed teleport can send a robot fifteen cells from where you expected. Spend the opening seconds mapping rather than moving.

Check which robots share a row or column with portal A and which direction they would travel to reach it. Those are the only robots whose slides you cannot predict by eye. Everything else on the board behaves exactly like standard play, which narrows the thinking considerably.

Then check whether portal A sits between a robot and the stopper it would otherwise reach. That is the classic trap: a move that looks like a two-cell nudge becomes a cross-board journey because A was quietly in the path.

Planning a double teleport. The second crossing does nothing and the robot runs on to the far wall, which usually means restarting the round.

Treating a portal as a destination. Robots never stop on portals, they pass through them. If you need a robot to rest near a portal you still need an ordinary stopper beside it.

Ignoring the exit side. A teleport that drops a robot into an empty lane with no stopper is worse than no teleport at all, because you have spent a move and finished further from the target than you started.

Reset freely while learning. Portal Mode has no timer and no leaderboard, and the fastest way to internalise a portal's geometry is to let one wrong slide play out completely and watch where the robot actually lands.