Report a bug
The Catalyst
One slot in the middle, and everything runs through it. Load it from your hand or off the ring itself, drag out to pick what it should join, and let go — where you let go says what you want done with the coin. Same Act One board as First Light and The Arm.
The Catalyst · prototype 71
Every verb in the game is now one gesture, and which one it is depends only on where your finger ends up.
Four coins along the bottom. Tap one and it loads into the core already reactive — no separate heating step, it is hot the moment it is in there, and you can go straight for a join. The supply is no longer random; it is a choice among four.
Nothing ever loads itself. After a spend the core is empty and stays empty until you pick again. Loading is a decision every single turn, which is the whole point of having a hand at all.
The queue outranks the ring. Press a queue coin and it takes the slot, sending any ring coin that was loaded back where it came from. A ring coin dragged at an occupied slot bounces — it flies home and the loaded coin flashes to say why.
So once you press queue you are committed to spending a queue coin before you can do ring work again — with one mercy: tap the loaded coin a second time and it goes back, as long as you have not dragged it yet. You have not acted, so nothing is owed. Once you drag, you are in.
A slot only refills when you spend it. A hand of four things you do not want is a squeeze: you must play one to see a new one, and playing one you do not need is exactly how the ring fills up.
The hand never shows the same kind twice. Not on the deal, not on a refill, not after a death. If it cannot fill four slots with four different things it shows fewer slots rather than padding with copies — so the Planck epoch has one button, inflation has two, and from the quark epoch on it is four. Four identical cards were never a choice, they were one choice drawn four times, and the deck now says so.
Spend a kind that still has work to do and the same kind comes back. If you need a third photon, the game stops making you fish for it. The slot only turns over when its coin has nothing left on the board to pair with.
And the draw leans towards what helps you build. A coin that can make something with what is on the ring is three times as likely as one that cannot — counting only recipes that build. Annihilation still clusters, brackets and plays; it just never gets pushed at you, because the board offering you the thing that destroys your quarks is not help.
That had been quietly broken. The ladder test used each item's position in the recipe table, and photons are produced by recipes that sit late in it — so annihilation scored as a step up and antimatter was being pushed at you after all. A recipe now counts as building if it returns a single coin, which is exact. Fixing it roughly halved how often annihilation clusters form.
Together these move the difficulty off did the game give me the right coin and onto can I manage my space while I assemble. The hand is generous on purpose; the ring is where it costs you.
That is the whole pressure system now. Every coin on the board is there because you put it there, and overflow is your own doing rather than a bad draw.
Eight outcomes, one sentence: where you let go says what you want done with the coin — except a queue coin, which is new material and always ends up on the board.
It reacts wherever you are. If the set is a recipe, letting go at the middle works as well as letting go at the rim. The middle is the commit button. Release at the rim and the new coin appears there; release at the middle and it appears where the first coin you picked was standing.
A failed queue coin lands on the ring — by the coins you picked if you were at the middle, under your finger if you were at the rim. It is new material and it has to go somewhere.
A failed ring coin goes back where you put it. Let go at the middle and it stays loaded, still hot, ready to try again. Let go at the rim and it lands there. Experimenting with what you already have is free, and should be.
Either way the core and the coin shake, so a failure is never silent.
Grab a coin and drag it round the ring and it ploughs everything in front of it. This is how you organise: sweep a region clear, herd the things you want next to each other, open a gap.
Let go anywhere and it wedges in, forcing its neighbours apart to make room. Moving never reacts, so the ring is a workspace, not a minefield.
Pull a coin off the ring and into the core. It sits there glowing red. That is the whole of "catalysing" — the core has exactly one slot, and while a coin is in it no new material can come out.
That is the real cost of heating something. The core is also your supply line, and parking a coin in it blocks the supply until you use it or take it back.
Nothing on this board ever merges by itself. There is one line of code that can cause a reaction and it runs when you let go of the wand — not on contact, not on a drag, not on a cascade. The old automatic matcher has been deleted outright rather than left dormant.
Heating a ring coin and picking from the hand put a coin in the same one slot, and it comes out the same way — while a heated ring coin sits there the hand is greyed out, because parking something in the catalyst still blocks your supply.
Drag it out and it is a wand. Take whatever you want it to react with — they do not have to be touching each other — and the core draws a thread to each one. Three, plus the one in your hand, makes four.
Taking one is a poke outward, never a slide along the rim. Cut across the empty middle to the angle you want, poke out at the coin, drop back in, cross to the next, poke out again. One coin per poke, and the rim lights up whenever it can take another.
This is what makes a tight cluster safe. Sliding your finger across three touching coins now takes nothing — only a deliberate move outward counts. Poke the same coin a second time and it lets go again, so a wrong pick costs a flick of the thumb.
While a finger is down the board answers to the finger and nothing else. No drifting together, no re-spacing, no cluster chasing your hand. Lift one member of a cluster and the rest stay exactly where they were, to the decimal.
The one exception is the shove, which has to push coins aside — but even then the board does not reorganise itself until you let go.
The opening board does not move at all until you touch it. Seven identical super-forces all attract each other, so the first thing you used to see was the game rearranging itself before you had done anything. Now it waits.
A red cord runs from the core to the coin the whole time it is out, so it always reads as something drawn from the reactor rather than a loose piece.
Let go beside one of them. The reaction happens where your finger is, and the new coin appears right there — it never flies to the far side of the ring to fuse. Release too far from every ingredient and it shakes back to the core, unspent.
Otherwise: it fuses and the threads snap, or it shakes back and nothing is lost. A failed reaction costs a gesture, not a life.
To cool a coin down again, sweep it to an empty stretch of ring and drop it there.
Coins no longer hold a fixed spot. Left alone they spread themselves evenly around the ring, the way Atomas does — the board always looks composed, and gaps close themselves.
The exception is the whole point: coins attract each other only if they can react. Partners snap together into a cluster, joined by a green bracket outside the rim. Everything else politely spreads out.
But the reach is short — about two coin-widths. Two partners sitting across the board from each other do nothing at all. You have to shove them together, and when they are close enough they snap and the bracket appears. The board does not hand you a map; it confirms a guess.
That is what makes the shove the third real verb rather than tidying. Clearing a stranger out from between two coins is now a move, because a stranger in the middle breaks the cluster outright.
The reach scales with coin size, so it means the same thing in act nine with tiny coins as it does in act one with big ones.
A cluster can never grow past one recipe. Two super-forces make gravity and three make the strong force, so three will gather — and the fourth is pushed away, because no recipe takes four. Four photons do stay together, because four photons are a recipe.
The bracket says three different things:
Solid green — a whole recipe that builds. Merging it puts something new on the board and takes coins off it. Two super-forces into gravity, three into the strong force, two ups and a down into a proton.
Solid yellow — a whole recipe that breaks. The three annihilations hand you two photons back for two particles. It is a real move, and sometimes the move you want when the ring is drowning — but it is buying room with material rather than making progress, and the colour says so before you commit.
Dotted red — only part of a recipe. Two energies when it wants three. It will not join.
Size has nothing to do with the colour. Two super-forces and three super-forces are both builds, so both are green. An earlier version marked the pair green and the triple yellow, which was an accident of the test rather than a statement about the board — the rule looked for a smaller recipe hiding inside the cluster, and a pair has nothing inside it to find.
We tried the stricter reading first — yellow whenever these coins have a second use somewhere on the ring — and measured it across fifteen hundred boards an epoch: nought to four per cent green. On a real board almost every coin has a second use, so that rule collapses back into solid-versus-dotted and tells you nothing. Build-versus-break is the one distinction that stays informative once the ring is busy.
Nothing unrelated may sit inside a cluster either. Partners with a stranger between them stay apart until you shove the stranger out — which is what the shove is for.
Attraction is also suspended for as long as you are touching the board. Otherwise a cluster follows the coin you are dragging, which is exactly the sort of thing that makes a board feel possessed.
So the board tells you where the opportunities are without telling you what they make. A nudge, not an answer.
Dropping the adjacency requirement is the big one. Arranging four specific things into physical contact on a crowded ring was the least fun part of every version so far — fiddly, and getting worse as coins got bigger.
The rim rule fixes the other half. A sweep used to grab whatever it brushed past, so a photon sitting between your proton and your electron would wreck the reaction. Now the interior is a corridor and only the rim is live.
Your line was the one that made it click: the centre is our catalyst. Once it is the hot place, every gesture explains itself — the ring is cold storage, the middle is where things react, and you bring ingredients toward the heat.
It also happens to be true. Nucleosynthesis is a heat problem. A star is a hot centre with cold material falling toward it. The mechanic and the physics are the same shape, which is rare and worth keeping.
The core has somewhere to go, too: the singularity, then a stellar core, then a cell nucleus, then a reactor, then a mind. Same object, redrawn once an act.
Your thumb is on the board. The ring is bigger and the coins smaller now, and the grab target never falls below a thumb's width — but picking one photon out of a crowd one-handed is still the thing to test first.
Chain reactions are gone. Every merge is one deliberate gesture — your "nothing auto-merges" rule honoured to the letter — but it costs the cascading run of merges that most of the genre lives on.
Even spreading means the ring never stays where you put it. Ploughing a coin somewhere only holds if it has a partner to cling to — which is now the point of ploughing rather than a side effect of it.
On a nearly empty ring — three or four coins — nothing is close enough to cluster and no brackets appear. Harmless, but worth knowing it is the physics and not a bug.
A poke also costs more motion than a brush. Gathering three ingredients is out-in-out-in-out rather than one smooth arc — more deliberate, and slower. That is the trade for never taking a coin you did not mean to.
The ring overflowing is the only way to lose, and it costs a life. What happens then: the most basic coins burn off, from the bottom of the ladder up, until the ring is back to about 40%. Not a fixed number kept — a fixed amount of room made.
So you lose raw material, never what you climbed to. A board of thirty super-forces loses eighteen of them. A board with a proton, a deuterium and a helium-4 on it keeps all three and burns the hydrogen around them. The thing you spent ten minutes building is the last thing to go.
Every discovery is kept regardless — the codex never rolls back.
Whatever was loaded in the core goes back on the board first, so it is judged on the same ladder as everything else rather than just vanishing.
A death always hands you a free hint, and it does not cost one of your three. If you died because you could not see the move, you will see it now.
And you get a fresh hand, dealt against what survived. The old one was drawn for a board that no longer exists. Four different kinds, every one of them able to work with something still standing — so you come back with a move to make rather than a hand of material for a board that just burned.
Dying takes its time. The board freezes where it stands, the doomed coins swell red and burn out over about a second and a half, the ring fill visibly drops as they go, and then a card comes up: what burned, what is left, and how many lives you have. Nothing moves again until you tap it.
The free hint is on that card rather than flashing past in the banner, so the thing that explains the death is part of the death. Tap to carry on and it stays in the banner while you work.
If you die often you can tap during the burn to skip to the card.
No sentence to parse. The hint draws the recipe: the coins that go in, the ones you are missing shown as empty dashed outlines, and what it makes, ringed in green.
It stays up until you tap. Nothing times out from under you mid-read, and a tap anywhere on the phone clears it.
The same picture appears on the death card, so the free hint you get for dying reads exactly like the ones you pay for.
3 lives, 3 hints, both refilled at every epoch gate — so you always start a hard stretch full.
Overflow costs a life. The basics burn off until the ring has room again; every discovery stays in the book. Cost, not setback.
Each death spends nothing and tells you something: the hint fires automatically, free, naming what you were closest to.
Top-left, in gold. It starts before zero and moves as you discover — 10−43 seconds, then 10−32, then a whole second, then three minutes. Temperature falls beside it.
By the end of Act One you have been playing for maybe twenty minutes and the universe is twenty minutes old. That alignment is worth protecting.
The ring widens and the coins shrink every epoch, so the board holds more as the game gets more complicated. Twenty-eight protons fit at the Planck epoch; fifty-eight by nucleosynthesis.
Both changes ease in over a second or two at each gate, which makes the epoch turn feel like the universe expanding rather than a settings change.
Coins never shrink below a floor, and the grab target never shrinks at all — it stays at least a thumb wide however small the coin is drawn.
The strip above the ring is not just a legend now — tap a coin and it opens. What the thing actually is, in one line of real physics. How many are on the ring. What it pairs with, of the things in front of you right now. And every recipe you have already made with it.
The spoiler line is drawn carefully: it names partners that are on your board, which the green brackets already hint at, and recipes you have already discovered, which is memory rather than a hint. It never names a recipe you have not found.
Tap anywhere to close it — the card, the board below it, the hand, the bar. Nothing needs aiming at.
It is also where the writing lives. "A free neutron decays in fifteen minutes, so use it" teaches the mechanic and the physics in the same breath, which is the only kind of tutorial this game should have.
The strip above the ring lists every distinct thing currently on the board, with its name. It is the board's legend, not the codex — it shrinks as you fuse and grows as you discover. It matters more now that small coins carry no glyph.
In the bar above it, ♥♥♥ are lives, lost to overflow, and ??? are hints you can spend, with the ? button to spend one. Both refill at every epoch gate.
The bottom of the screen is the hand and nothing else — no legend, no verb buttons. The three gestures are taught once on the opening card.
Twenty-four recipes. The gate item of each epoch is what carries you into the next one — make it and the era turns over.
✦+✦ = ⊙ gravity
✦+✦+✦ = S strong force
S+⊙ = EW electroweak
EW+EW = Ⅰ inflaton
One super-force, breaking apart as it cools. Gravity leaves first — it really did.
δ+δ = E energy
δ+δ+δ = ⊚ density seed
E+E+E = ≋ quark-gluon plasma
⊚+⊚ = ◍ filament
The ring physically grows here. Space expanding is the one moment where more room is the reward.
≋+E = u up / d down
E+E+E+E = e⁻ electron
u+ū = γ+γ annihilation
u+u+d = p proton
Antimatter arrives in the spawn pool. Annihilation is the pressure valve: two beads become two small ones, and a full ring gets survivable.
u+d+d = n neutron
e⁻+e⁺ = γ+γ
p+n = D deuterium
Matter beats antimatter by one part in a billion. Everything you build after this is that remainder.
D+n = T tritium
D+p = ³He
D+D = ⁴He
⁴He+T = ⁷Li
p+e⁻ = H
H+H = H₂ — end of act
The universe is a fusion reactor with three minutes on the clock. Then hydrogen and helium and nothing else for a hundred million years.
Once the clock passes a few hundred thousand years it can stop being physics and start being dates — 3.7 billion years ago, 66 million, 1969, and then announced targets for the near future: a crewed Mars attempt, a self-sustaining settlement, the first off-world manufacturing.
Those last ones are real published targets and they will slip, which is fine — we show them as targets, not history, and we name missions rather than people, so the game never puts words in a living person's mouth.
Can you reliably grab the coin you meant, one-handed, on a full ring? That is the only question this scheme can lose on.
Does ploughing feel good, or does it feel like the board is fighting you? It should feel like sweeping marbles with the back of your hand.
Is poke-out-poke-in too much work for a four-ingredient recipe, or does the certainty pay for itself?
Is four the right ceiling? Three would bite harder; five would almost always have what you want, which is no pressure at all. And does the one-button Planck deck read as honest or as broken — is it obvious that there is only one material to have?
Difficulty is now entirely your hit rate. A success eats two to four coins and returns one; a failure adds one. The ring shrinks when you are right and grows when you are wrong, and there is no way to stall. That is the whole balance, and it is worth knowing that is what we chose.
Dragging a coin to the middle and holding it there does nothing yet. That gesture is deliberately left empty — it is the obvious place to put a second verb if the loop gets thin after an hour.
Is the bracket too generous now that it only ever marks real partial recipes? In a late act with hundreds of items it may still mean brackets everywhere — the fallback is to show it only for sets that are one coin from complete.
And is one core slot enough, or does blocking the supply while a coin is heated bite too hard early on?
The Catalyst — counting…. Every number here is counted out of the game's own tables when this page loads, so it cannot go stale.
'+EP.length+' where it meant the number of epochs — found by looking at a screenshot, not by a test; and this list brought up to date.