transcript
0 nt left
A codon translates the instant three bases touch. Same-class residues bond on contact.
0 fragments
longest chain — / 12 to win
no grid — translation happens on contact
No cells, no lattice, no fixed machine. Bases are beads in a vessel, and the moment three of them are touching each other, they translate — right there, wherever that happens. The resulting residue is done with nucleotides for good, but it bonds to any residue of its own class it touches, and a growing chain is a literal cluster of linked beads that sprawls as it grows, not a line. Every drop is yours — nothing arrives on its own. The vessel still fills, because a peptide cluster is genuinely bulky and only a release factor reliably clears one; how big you let a chain grow before you cash it in is the trade the mode runs on.
Bases are beads. A fragment of two or three is held rigid by distance constraints, so it tumbles as a unit and comes to rest at whatever angle the solver gives it. There is no ribosome to carry a codon to — the moment three nucleotide beads are mutually touching, they translate right there, in place.
A dropped 3-base fragment satisfies this the instant it lands, in whatever order rotation gave it — that is the reliable, controlled play: your whole codon, your chosen order. A single or a pair of loose bases satisfies it opportunistically, whenever it happens to end up touching two others already in the pile — the oldest of any three touching bases reads as the 5′ end. That is the chaotic play: you are seeding the pile, not dictating the outcome.
Once translated, a residue never re-enters codon detection and no nucleotide ever bonds to it — the two chemistries do not interact. What a residue does do is bond to any other residue of its own biochemical class it happens to be touching. There is no designated chain end to append to: a new residue bonds wherever contact puts it, so a growing peptide is a real, sprawling cluster of linked beads — the shape emerges from where things actually collided, not from a line filling up. Two established clusters that bump into each other merge the same way.
This is the third design. The first had a fixed ribosome the strand was pulled through; the second still had that ribosome, just accelerating the feed that outran it. Anchoring translation to contact instead of to a place removes the machine entirely, and it is what makes merging automatic rather than a click.
Nothing arrives on its own. You choose when to release the next fragment and where, with as much time as you want to look at the pile first. An earlier build had fragments arrive on an accelerating clock whether you wanted them or not — it made the vessel fill reliably, but it felt like being rushed by the game rather than working the pile, so it is gone.
The vessel still has a line on it that the pile must not reach, and it is still reachable: a peptide cluster is genuinely bulky — several beads wide, not a compact number — and a release factor, the reliable way to clear one, only arrives once every 12 codons. Over a full run that is a handful of charges against a great deal more material than that, so some of what you build will sit in the vessel taking up room no matter how carefully you play. How big you let a chain grow before you spend a charge on it is the trade that replaced the clock — bigger banks for more, at the cost of the room it was taking up in the meantime.
Measured: even a patient bot that considers every position before every single drop does not avoid this reliably — median peak fill still reaches about half the line, and a real fraction of runs top out regardless. Where you drop barely moves that number, which is consistent with every version of this mode so far: this vessel does not reward spatial aim. What does move it is which residue you choose to make — rotating a fragment to match a class you are already growing measurably outscores rotating at random.
A burst of nearby translations — a fragment landing here, a loose match completing there — keeps a shared multiplier alive at ×1.5, then ×2, then ×2.5 and upward. Let the pile go quiet and it drops back to one.
A stop codon becomes an inert marker that drifts through the pile; the instant it touches a chain of two or more, that chain banks and the marker is consumed with it. Or spend a release factor, earned one per 12 codons, by clicking any residue in a chain — on the canvas, or in the list below it. Banking pays length² × class rarity.
The pile reaches the line. There is no other way to lose, and it is reachable even under careful play — see above.
The other three modes tell you exactly what a move will do before you commit to it. This one only promises the physics is deterministic — a pile of round beads is not predictable, and that is the trade this mode makes. If you want to know the outcome of a placement before you make it, play Tumble.