The four-round arc

You get 26 turns. Ever. Eight in round 1, then 7, then 6, then 5. At the end of each round you spend one of your eight action cubes marking the goal board, which is why the count shrinks.

Look at the shape of that: rounds 1 and 2 are 15 of your 26 turns, 58% of the game. Rounds 3 and 4 are 11. The game front-loads your actions and back-loads your scoring, and that is the whole strategic problem. You have the most turns when your engine is weakest, and the strongest engine when you have the fewest turns left.

Early turns are worth more, literally

A bird played on turn 1 can have its row activated up to 25 more times. A bird played on the first turn of round 4 can be activated at most four more times.

Same card, different card. A rough guide to how many times a brown power in your main row will still fire, a modelled rule of thumb rather than measured data:

Played in	Times it still fires
Round 1	~7-9
Round 2	~5-6
Round 3	~3-4
Round 4	~0-1

A power worth one point per activation is worth 7 to 9 points played on turn one and almost nothing played in round 4. Late birds have to justify themselves on printed points, not on their power. That is the whole reason judging a bird depends on the clock.

Rounds 1-2: build (15 turns)

One job: get your rows off their base slots and get brown powers online.

The base actions start bad on purpose. An empty forest gives 1 food per cube. An empty wetland gives 1 card. The benchmark to reach is two food per forest action and two cards per wetland action, which means two birds in that row.

• Cheap brown-power birds beat expensive high-point birds here. A 2-point bird that gains a food on every forest run out-earns a 6-point bird played on turn 3, because it fires all game.
• Prefer worm and seed costs. Each is on 33% of die faces. Fish, fruit and rodent are 17% each, so those birds strand you.
• Tempo marker, an estimate rather than data: 3 to 5 birds down by the end of round 1, 8 or 9 by the end of round 2, with at least one row holding two birds.

Round 2 is the last round where a new engine piece pays for itself. After that you are buying a machine with no time left to run it.

Round 3: the pivot (6 turns)

Not a switch you flip. A test you run before every cube.

The pivot test. If a new engine piece cannot pay for itself before the round ends, stop buying engine and start buying points.

Do the arithmetic out loud, three ways:

• Run the grassland with four birds in it: 4 eggs this turn, repeatable, before any brown powers add more.
• Play a 5-point bird into column 4 or 5: 5 points once, minus the 2 eggs the column costs, so 3 net, for the same single cube.
• Play a 2-point bird with a real brown power: 2 points now, plus three or four more activations of that power.

A useful smell test on expensive birds is the 3/6/9 scale: one food spent should return 3 or more points, two food 6, three food 9. A bird that fails that in round 3 is usually worse than just running your best row again.

Round 3 is normally mixed. Play the two or three point birds you banked food for, and run the engine with everything else.

Round 4: harvest (5 turns)

Almost everything in front of you is about to be worth zero.

At scoring	Worth
Egg on a bird	1
Tucked card	1
Food cached on a card	1
Food in your supply	0
Card in your hand	0

Five cubes, all conversion. Do not gain food unless you spend it this round. Do not draw cards unless a tucking power turns them into points or you already know which bird you are playing. Count your open egg spaces before committing to a grassland run, because excess eggs beyond a bird’s limit are lost.

The full last-round checklist lives on The endgame.

The grade you give each round

Ask one question at every round boundary: did this round make the next one richer? Rounds 1 and 2 should answer yes in resources. Round 3 should answer yes in points. Round 4 has no next round, which is exactly why you play it differently.