Chain Mode: Winning When Every Target Must Hold at Once

In standard play a round ends the moment the target robot touches the target cell. Chain Mode removes that instant win. Several coloured targets sit on the board, each bound to a robot of the same colour, and the round ends only when every one of those robots is resting on its matching cell at the same time.

That single change makes the puzzle non-sequential. Reaching target one, then target two, then target three is not progress if the third solution knocked the first robot loose. What you are actually searching for is a final board state, together with a move order that arrives at it without destroying itself along the way.

Players arriving from standard mode usually lose their first several Chain rounds to exactly this misunderstanding. They treat each individual arrival as a checkpoint, then discover on the last move that the position they built cannot be completed. Unlearning the checkpoint habit is most of the early progress in this mode.

The practical consequence is an inverted order of operations. Identify which robot is hardest to land, usually the one whose target has the fewest natural stoppers, and treat its solution as fixed. Everything else gets arranged around it.

Then work outward. For each remaining robot ask not only how it reaches its own target, but whether that route crosses a cell the hard robot needs, either as a travel lane or as a stopper. If it does, that robot has to move first, before the hard robot commits.

In effect you are sorting the robots by dependency. Anything that must be out of the way goes early, anything that will serve as a permanent stopper goes early, and the robot with the most fragile solution goes last. A surprising number of Chain rounds that look impossible are simply being attempted in the wrong order.

The most expensive mistake in Chain Mode is cascading displacement. You slide blue across the board toward its target and on the way it collides with green, which was already resting on the green target, and a solved position quietly becomes unsolved.

Robots never push each other here; the moving robot stops. In reverse, that is precisely the difficulty. A robot resting on a target is a wall for everyone else, which makes every solved target simultaneously an asset and an obstacle. It is a free stopper you can plan around, and a lane you can no longer travel.

Before committing any move, trace its full slide and name the object it will stop against. Half of Chain Mode skill is noticing that your own solved robots have turned the board into a maze that did not exist when the round started.

Reverse search still works, but you run it once per target rather than once per board. For each target, list the directions a robot could arrive from and what would have to stop it there. Do this for every target before planning a single move.

Then compare the lists. Where two targets need the same cell occupied you have a conflict to resolve. Where one target's required stopper turns out to be another target's robot sitting in its final resting place, you have found the round's key: those two solutions reinforce each other and should be built as a pair.

This is why Chain boards tend to have one clean solution rather than many. The constraints interlock, so the work is less about finding a path than about finding the single arrangement in which no constraint contradicts another.

Start with two-target rounds and refuse to move until you can state the final cell of both robots and which object stops each of them. If you cannot state it, you are guessing, and guessing scales badly as targets are added.

At three and four targets add one more discipline. After every move, re-scan the solved robots and confirm that none of them is now blocking a lane you still need. It takes a few seconds and it prevents the reset that costs a minute.

There is no timer and no leaderboard in Chain Mode, so treat a failed attempt as information rather than a loss. The position that trapped you usually reveals which robot should have moved first, and that ordering insight transfers directly to the next board.