Grid Network: a strategy guide
Grid Network is a city-planning puzzle in miniature. A power plant sits somewhere on a 3D landscape of grass, water, hills and rock, and a handful of houses and factories need connecting to it. Every tile you route a cable through costs something, and the terrain is what decides how much.
It is a friendly disguise for a genuinely deep problem — the same shortest-path and shared-infrastructure trade-offs that real utility planners face. The scoring compares your total against the mathematically cheapest network, which is computed with a proper shortest-path search rather than a rough estimate.
The rules, restated
- Every building must end up connected to the power plant by an unbroken cable run.
- Cables are laid tile by tile. Each terrain type has its own cost — grass is cheap, water and rock are expensive.
- Cables may share tiles. A shared run is paid for once, which is where most of the savings come from.
- Each connection is drawn as its own coloured cable, and cables sharing a tile edge spread into parallel lanes so you can always see which line goes where.
- Undo steps back one tile at a time; clear wipes the board.
- Your score is your total cost measured against the optimal network for that level.
A worked example
Picture a plant on the left, two houses on the right, and a river of water tiles running down the middle. Water costs several times what grass costs.
The naive solution runs a separate cable from the plant to each house, and each one crosses the river independently. That means paying the river crossing twice.
The cheap solution runs a single trunk line from the plant to one crossing point, over the river once, and then branches on the far bank to reach both houses. Even if the trunk is longer in tiles, it is dramatically cheaper because the expensive tiles are shared.
This is the whole game in one picture: find the expensive obstacle, cross it as few times as possible, and branch on the far side.
Cost, not distance
The single habit that separates a good Grid Network score from a mediocre one is counting in cost rather than tiles. The 3D board makes distance visually obvious and cost visually subtle, so the eye keeps proposing straight lines through terrain that is quietly charging a premium.
A useful exercise on any level: before laying anything, mentally price the straight-line route from the plant to the furthest building. Then price a route that curves around the expensive band. On most levels the curve wins by a wide margin, and once you have seen that on one level you stop trusting straight lines.
This also reframes what a good crossing point is. You are not looking for the narrowest part of the river in tiles; you are looking for the cheapest crossing overall, which might be a slightly wider stretch of cheap water rather than a narrow stretch of rock.
Trunk-and-branch thinking
Because shared tiles are only billed once, the optimal network is essentially a tree: one expensive spine, with cheap branches hanging off it near each destination. Levels are designed so the naive star layout — one cable per building, radiating from the plant — is always beatable.
The practical method is to build the most expensive single connection first, treating it as the spine, and then connect every other building to the nearest point on that spine rather than back to the plant. That one change typically closes most of the gap to optimal.
The refinement after that is to shift the spine. Moving the trunk two tiles sideways can shorten three branches at once. This is where the last few percent of score lives, and it is the reason undo exists.
Common mistakes
Connecting buildings in the order they catch your eye. The nearest building is the worst one to start with, because a short cheap cable tells you nothing about where the trunk should run.
Crossing expensive terrain more than once. Every extra crossing is usually worth more than any routing cleverness elsewhere on the board.
Stopping at the first valid network. Valid and cheap are different things, and the score screen compares you to the optimum, not to whether the lights came on.
Quick reference
Find the chokepoint first — Every level has one stretch of expensive terrain that separates the plant from most of the buildings. Decide where you cross it before you lay a single tile.
Build trunks, not spokes — Separate cables from the plant to each building is almost always the most expensive layout possible. Share as much of the route as you can.
Detours are often cheaper — A route that is four tiles longer but avoids two rock tiles usually wins. Count cost, not distance.
Route the far building first — The longest connection determines where the trunk should go. Build it first, then hang the nearer buildings off it.
Undo is free — There is no penalty for experimenting. Lay a candidate trunk, look at the total, undo, and try the other crossing.
Frequently asked questions
- How is the optimal cost calculated?
- The level's cheapest network is computed with a shortest-path search over the terrain costs, so the target you are scored against is genuinely optimal, not a guess.
- Can cables cross each other?
- Yes, and sharing tiles is encouraged — a shared tile is only paid for once. Overlapping runs are drawn in parallel lanes so the picture stays readable.
- Is the 3D view required?
- The board is rendered in 3D because terrain height carries information, but every interaction is a simple tap or click on a tile.
- When is Grid Network the featured game?
- It takes the Game of the Month slot every third month, and stays playable as practice in between.