Strata: how a small mining game became 1,400 metres deep
It began as a vertical slice about the feel of digging. Nine iterations later it's the only game on this site that saves your progress, hands you a week-long objective, and rebuilds the world every Monday.
Every other game we've made is a run: three to ten minutes, a score, a leaderboard, done. Strata broke that pattern, not because we set out to build something bigger but because each thing we fixed made the next limitation obvious.
The slice: does digging feel good?
The first build did one thing. You could dig, in four directions, and the tiles took time to break depending on what they were made of. No economy, no upgrades, no objective. Just a hole and the feel of making it.
That was the right thing to build first, because the answer wasn't obvious. Digging downward is satisfying in a lot of games; digging sideways is usually pointless, and we wanted a world that was as wide as it was deep. The slice was there to find out whether cutting a horizontal tunnel felt like exploration or like wasting time.
It felt like exploration, but only once ore came in clusters rather than being scattered evenly. A single ore tile is a chore. A seam you follow sideways for twenty metres, wondering when it stops, is a decision.
The loop: haul, sell, upgrade, go deeper
Next came the economy, and with it the first real design constraint. Once you can spend credits on a better drill, the whole game becomes a question of what the player buys and in what order — and if any one upgrade is clearly correct first, there's no decision at all.
Six tracks: drill, cargo, fuel, lamp, jetpack and hull. What makes them interesting is that they gate different things. The drill gates how hard the rock can be. Hull gates how deep you can survive. Cargo gates how much a trip is worth. Buying only drill tiers gets you into rock you can cut but not into depths you can live at.
The jetpack is the one we're happiest with, because tier zero is nothing at all — you can only jump. Buying the first tier is the single biggest change in the game, because vertical movement stops requiring a staircase. It converts the world from something you tunnel through into something you fly around in.
The world: hazards that make depth mean something
A deeper hole isn't automatically more interesting. Depth had to start costing something other than travel time, so each stratum brought a hazard that changes how you dig rather than just how much damage you take.
- Water pools in cavities and halves your speed while you're in it — harmless, but it turns an escape into a crawl.
- Gravel falls when you remove what's under it, so you have to think about what's above the tile you're standing in.
- Gas pockets vent fuel the instant you breach one. Not damaging; just a long way from home with nothing in the tank.
- Lava burns through hull fast, and cutting into a pocket from below is still the most common way a good run ends.
Only the last of those does damage. The other three attack your ability to leave, which we've found is consistently more frightening than attacking your health.
The problem we made for ourselves
Then we raised the world from 520 rows to 1,400, and immediately broke the game.
At 520 metres, restarting at the surface is a mild inconvenience. At 1,400 it's unplayable — every session begins with ten minutes of commuting through tunnels you already cut, to reach the place where the game actually is. We'd made the world deep enough that its own depth was the main obstacle.
The fix is a station every 100 metres down the main shaft. Reach one and it's banked permanently; from then on you can drop straight to the deepest one you've touched. The rule that makes it work is what stations don't do: they never skip a descent you haven't already made yourself, and they don't help on the way back up carrying something. They remove the commute and leave the expedition.
The objective: five Seals and a very long walk
A world that big needs a reason to cross it, and "get rich" stops motivating around the point where you can afford everything.
So: five Seals, one per stratum, from 48 metres down to 1,105. Each sits inside a vault chamber at a known depth but an unknown column, which is what finally justified all that horizontal space — finding one means tunnelling sideways at depth and hunting. Hold all five and the Core Vault at 1,330 opens.
What's inside has to be carried back to the surface by hand. No station helps. That walk is the whole game arguing with you, and it's the part we most enjoyed building.
The last change: everyone digs the same mine
The final iteration was the one that changed the game's character most, and it started as a leaderboard complaint. With a private random world per player, two people were never attempting the same thing — one gets a seam near the shaft, one gets forty metres of empty clay, and the board ranks luck.
Now a single world is generated every Monday at 00:00 UTC and everybody digs it. The seed comes from the ISO week, so no coordination is needed and rotation happens by itself. We've written that up properly in a guide about building a fair procedural world, because it turned out to matter far beyond this one game.
The side effect we didn't plan for: it made Strata social without a single social feature. Everyone is working on the same puzzle in the same week, so "has anyone found the Granite Seal yet" is a question that means something.
What we'd do differently
Build the stations before deepening the world, not after. The order we did it in meant several days where the game was, in practice, unplayable past a certain depth — and it's very hard to judge whether a hazard is well-tuned when reaching it takes ten minutes.
And write the page at the same time as the game. That one cost us more than we expected, and it has its own post.