Planning and mechanism puzzles: thinking a few moves ahead
Many puzzles ask you to look ahead, but they ask in two quite different ways. In one kind you choose a sequence of moves and have to find a short one among many. In the other you place a few pieces, press go, and a chain reaction runs without you, so the whole job is predicting what it will do.
We call these planning puzzles and mechanism puzzles. This guide explains how each one is solved, what par means, and walks through small examples taken from the actual rules of eight of our daily games.
Planning: searching for a short route
In a planning puzzle every move is yours, and the question is how few you need. Pipstep, Tiltwood and Porchlight work this way. The honest method is a small search: try a line, see where it leads, step back, try another. Undo is free in all three, so exploring costs nothing.
What turns a blind search into a smart one is pruning, which means throwing away whole families of moves because you can see they cannot work. Working backwards from the goal is the strongest pruning tool there is, and it comes up again and again below.
Par: a target, not a verdict
Almost every game here shows a counter with your moves next to par, the fewest possible. Matching par earns three stars. In Pipstep, up to three over par is two stars. In Tickwork and Pourwise it is one or two over. Porchlight routes always have an even number of moves, so the step after par is par plus two.
Par is useful because it is a fixed number you can aim at. If the counter says par is 7, you know a 7-move solution exists, so a 9-move one is not the best available and it is worth hunting for the shortcut. At par the day ends by itself; above it a Finish button appears, and you can hand in or undo and keep trying.
Three planning examples
Pipstep is a die rolling over a grid. Opposite faces add up to 7, so rolling towards a face brings up its opposite: if the 3 is on the right and you roll right, the 4 comes up. A straight line only cycles four faces, and the two on the sides never reach the top. That is where detours earn their keep: step aside, roll on and step back, and you spend two extra rolls to arrive with a different face up. So for each chip you work backwards: which face must be on top when I land here?
Tiltwood never lets you touch the marbles. You turn the whole board a quarter turn and everything falls towards the bottom until it hits the frame, a wall or another piece. A marble only stops against something, so look along the hole's row and column for a stopper just beyond it. That is where the marble must be one turn earlier. Boxes and other marbles are stoppers too, and a marble already home no longer blocks anything. Two turns the same way flip the maze upside down, while a left then a right moves every piece.
Porchlight gives the van a small tank. Every move burns one unit of fuel and a petrol station refills it as often as you like. The gauge is therefore a move count: if the full round is longer than the tank, the station is part of the route, and where you stop decides which parcels you can reach. Nearest first is rarely best, and a parcel down a dead end is cheapest to grab when you pass that corner anyway.
Mechanisms: predicting a chain reaction
In a mechanism puzzle, you set things up and the machinery does the rest. Tickwork, Pourwise and Onespark are examples. The skill is to simulate the chain in your head without running it, and, better, to find a quantity that does not change however you order things.
Tickwork has such an invariant. You push gears, each push turns the gear a quarter clockwise and its touching neighbours a quarter the other way, and the goal is every arrow up. Pushing A then B leaves exactly the same board as pushing B then A, so order is irrelevant: only how many times you push each gear matters, from 0 to 3, since four pushes make a full turn. Take three gears in a row pointing left, down, left. Push each end once. Each end gear turns a quarter clockwise and points up, and the middle gear, turned a quarter the other way twice, ends up pointing up too. Three pushes, with no search needed.
Onespark rewards counting before simulating. Bombs that blast each other form a group, and you usually need one extra bomb for each gap between groups, so counting the gaps gives a good guess at par. Pourwise rewards making differences: pour a full 8 into a 5 and what stays behind is 3, then 5 minus 3 is 2. To leave exactly 4 in the 5 jug you need it full with room for exactly 1 in another jug, so you start from the goal and ask what must happen just before.
Two special cases: Toywright and Ringcord
Toywright sits between the two ideas. You place machines on a belt in an order, and the order decides what covers what: paint hides stamps, the sander removes everything, and a second primary colour mixes instead of covering. So you start with the hardest order and work backwards. A stamp, for instance, always goes after the last paint. Colour sensors let one machine treat pieces differently, and on mixing days one paint can give two pieces two colours: yellow over blue makes green, yellow over red makes orange.
Ringcord is the odd one out. It is a 90-second shift at a village switchboard, scored against a perfect operator rather than against par. There is little to search ahead, but the same habits apply: unplug as soon as a call blinks red, look at urgent calls first, and learn where the icons are during the countdown.
Reviewing a run
Shared results show your stars, your moves and par, never the route, so if you want to learn from a run you need to remember it yourself. A good habit is to write down, after the fact, the one idea that decided the puzzle: the stopper behind the hole, the shared gear count, the refuelling stop.
The archive lets you play earlier days again. That is the best practice available, because par does not change and you already know the puzzle. Replay a day where you finished over par and try to reach it, or replay one you solved at par and see whether you can find the same line faster. Either way you are training the thing that matters: noticing which idea unlocks the board.