Strategy guides › Burst
How to last longer in Burst: prime first, fire big chains
▶ Play Burst🎯 Try the board from the diagram
Level 7, four drops left. One tap would set off a nice little chain of four. Satisfying pops, numbers flying. And when it ends you have three drops, because a chain of four pays back exactly as much as a chain of three, and that is one drop.
Burst is an economy game wearing a toy’s clothes. Every board costs more water than the one before, and the only income is long chains. We wrote bots that play our real game code and ran thousands of runs with each. The difference between a run that dies on level 6 and one that reaches level 9 is not luck. It is whether you are willing to make a tap that does nothing.
Every drop has to come back
The rules are simple. A tap costs one drop. A chain pays back one drop for every three bursts in it, rounded down, and clearing the board adds one more. So a tap that sets off nothing costs you one drop. Two bursts also cost one. Three bursts break even. Every burst beyond a multiple of three is spent for nothing: a chain of five pays one drop, the same as a chain of three.
That would be fine if the boards stayed the same, but they get worse in a very specific way. We generated 500 boards per level with the game’s own generator:
| Level | Bubbles | Size-3 share | Hits needed to clear |
|---|---|---|---|
| 1 | 18.1 | 45 % | 31.7 |
| 3 | 19.1 | 33 % | 37.7 |
| 5 | 20.2 | 21 % | 45.0 |
| 7 | 21.2 | 9 % | 52.5 |
| 9 | 22.4 | 0 % | 59.7 |
“Hits needed” is how much water the board is missing: a bubble of size 1 needs three hits to burst, a 3 needs one. Each hit comes from your tap or from a droplet. From level 9 on, the generator deals no size-3 bubbles at all. Whatever chain you fire there, you built yourself.
Size-3 bubbles in a line are the engine
A droplet flies straight over empty cells until it reaches the first bubble in its way. So what matters about a size-3 bubble is not what sits next to it but what it can see: the first bubble in each of the four directions. We took every size-3 bubble on 3,000 level-1 boards, counted how many size-3 bubbles it could see, tapped it, and recorded the chain:
| Size-3 bubbles in sight | Average chain | Chain of 3 or more |
|---|---|---|
| 0 | 1.0 | never |
| 1 | 8.9 | 75 % |
| 2 | 12.9 | always |
| 3 | 15.7 | always |
A size-3 bubble that sees no other 3 is a dud. It bursts alone, its four droplets nudge four bubbles up by one, and that is that. But a 3 that sees a single other 3 averages a chain of nearly nine, because the second burst sends droplets down new lines, and so on. When you scan a board, look for lines of 3s with gaps between them, not just clumps.
Pick the fuse, not just the cluster
Look at the cluster in the diagram. Tap the dashed bubble (row 5, column 4) and you get five bursts: the bubbles above, left and below it go together, one more pops at the bottom, and the chain stalls. Tap the green one right next to it (row 5, column 3) and you get twelve, clearing everything except four bubbles near the top and the 1 in the corner.
The difference is timing. When the green bubble goes first, its left droplet turns the 1 beside it into a 2, and the dashed bubble bursts one wave later. Its left droplet now flies through the cell the green bubble just emptied and hits that same bubble again. Three hits in all, so it bursts, and the chain runs up the left side. Light the dashed bubble first and the two go off in the same wave: the 1 gets hit twice, ends as a 3, and the chain dies there. The game is completely deterministic, so the same tap always does the same thing. Before you fire, try two or three fuses in your head and ask which small bubbles get hit twice.
Don’t fire small chains
Here is the habit that matters most. We compared three bots over 2,000 runs each, all playing the real game code:
| Bot | Average level | Reached level 8 | Best run |
|---|---|---|---|
| Greedy: always fire the biggest chain on the board | 5.8 | 16 % | 10 |
| Prime-then-fire: fire only chains of 3+, otherwise set one up | 7.4 | 53 % | 11 |
| Two-move lookahead: tries every pair of taps | 7.8 | 64 % | 12 |
The middle bot follows a rule simple enough for a human. If a tap clears the board, take it. Otherwise, if some chain pays at least one drop, fire the one that pays most, preferring the shortest such chain. If nothing pays, make a quiet tap: grow a 1 or a 2 that makes the biggest chain available next turn. Never fire a chain of one or two just because it is there.
Why it works shows in the bursts. The greedy bot throws away 36 % of its bursts to the rounding, all those fours and fives that pay like threes. The lookahead bot throws away 7 %, and 65 % of its taps burst nothing at all. It spends most of its drops quietly loading the board, then fires one long chain that pays everything back. The quiet tap feels like a waste. It is the opposite.
Spend early, but not on empty cells
Taps on empty cells are almost never worth it: the new bubble is a size 1 and needs three more hits before it does anything. And unused drops carry over, which matters because the levels get harder. A greedy run clears level 1 in 3.5 taps on average and level 8 in 13.7. The drops you save on the early, 3-rich boards are what pay for the late ones, where every 3 has to be built by hand.
Practice
The practice button opens seed 1006, whose first level is the board in the diagram. Find the twelve-burst fuse, then play the run with the prime-then-fire rule: a quiet tap is a drop invested, and a chain of two is a drop thrown away.
