Skip to content
Paul Kerrison
Go back

Carrier

Edit page
Carrier

Mutant Maker is a daily puzzle. You are shown a target mutant. It has a body colour, an eye count, a limb type, a pattern, and a size. You have four starting mutants. Breed them together, select the best offspring, breed again. Win when you produce a mutant that matches the target on all five genes.

That is the simple version. The interesting version involves Punnett squares, recessive alleles, inbreeding penalties, and a resource called Lab Power that runs out faster than you expect.

Why Genetics

The obvious way to build a game like this is to treat traits as probability slots. Set a percentage chance for each trait to inherit from each parent, roll the dice, done. It would work. It would also be boring.

Mutant Maker uses actual Mendelian inheritance. Each gene has two alleles: one expressed (dominant if one exists), one hidden. Some trait values are dominant. Cobalt body colour, legs for limbs, solid or stripes pattern. Some are recessive. Amber, Violet, Slate body colours, wings and tentacles, dots. When a mutant carries one dominant and one recessive allele, it shows the dominant trait but passes either allele to its offspring.

This matters because the target is biased toward recessive traits. Around three of the five target genes tend to be recessive. The starting population is seeded with carriers — mutants that look perfectly ordinary but are hiding the target value in their secondary allele. The game gives you a few log entries at the start to help: “Mutant 2 carries recessive Colour.” The challenge is getting those hidden alleles to surface and pair up.

How the Breeding Works

Each breed produces four offspring. Not randomly — the four offspring represent the four actual Punnett square combinations. If parent one carries alleles A and a, and parent two carries B and b, the four offspring get AB, Ab, aB, and ab. This guarantees exact Mendelian ratios: a cross between two carriers of a recessive trait produces one homozygous dominant, two carriers, and one homozygous recessive, in that order, every time.

There is a five percent chance per allele of a background mutation — a random value from anywhere in the gene pool. This is the fallback for situations where the target value genuinely does not exist in the starting population and you need a hail-mary. In practice it fires rarely enough to feel like luck rather than policy.

Before you breed, the Breed Preview panel shows you the Punnett square probabilities for the selected pair. For each of the five genes, it shows what phenotypes are possible and at what probability. This is the information the game is built around. Breeding without looking at it is guessing. Breeding with it is planning.

Lab Power and Genetic Stability

You start with one hundred Lab Power. Each breed costs ten. Force Mutation costs thirty-five per gene on top of that. You run out of power before you run out of patience, which is the point.

The Force Mutation panel lets you override a gene in all four offspring, guaranteeing the target value regardless of what the parents carry. It is the nuclear option. It costs so much that using it on more than one gene at a time means you will likely run out of power before you can win.

Genetic Stability starts at one hundred and drops each time you breed similar parents together. The drop scales with how genetically similar the two parents are: breed two siblings and stability falls hard. When stability gets low enough, deformities start appearing in offspring — random wrong values burned in regardless of parentage. You cannot recover stability once it falls. Breeding unrelated mutants keeps it healthy; hoarding your best mutant and treating it as the permanent parent of everything will eventually corrupt the line.

The tension between Lab Power and Stability creates most of the interesting decisions. You want to use the Gene Scanner to understand your population’s hidden alleles before committing to a breed. The scanner reveals the full genotype of a mutant: both alleles for all five genes. It costs nothing. Reading it before every breed and planning two or three generations ahead is how the puzzle is meant to be solved.

The Daily Puzzle

The same seeded PRNG approach used in This or That and Gold Fever appears again here. Date string to seed via djb2 hash, Mulberry32 for the sequence. The same date produces the same target and starting population for every player worldwide. Puzzle one is 29 April 2026.

State persists in localStorage. Close the tab, come back later, your generation and remaining power are waiting. If you re-open with less power than the minimum breed cost and no win on record, the game shows the fail screen. Closing a nearly-failed run doesn’t save it.

Google sign-in sits in the codebase via @react-oauth/google. The leaderboard infrastructure is there in outline. It has not done much yet.

The Stack

React 19, Vite, TypeScript, Tailwind 3. Framer Motion 12 for the scan animation and win screen. lucide-react for the gene icons. Zod is in the dependencies and is not currently doing much. No server-side logic, no database. The whole game is client-side with one exception: the OAuth flow, which currently authenticates users but does not persist anything for them.

RESULT

It’s live at mutantmaker.baudypositive.com. The genetics system holds up. The Punnett square preview is the best feature: understanding it is the gap between finishing in four generations and running out of power on generation seven.

The thing that took longest was the stability mechanic. The first version had no inbreeding penalty. Players would select the highest-scoring mutant each generation and breed it with itself, which the game allowed and which made it trivially solved. Adding stability gave the population shape: you need diversity, not just the best individual. Genetics turns out to have opinions about this.


Edit page
Share this post:

Previous Post
The SFW Version
Next Post
I Built the Greatest Prize Wheel Ever Made