Long reads
Longer pieces, written to be read rather than referred to. No prior knowledge assumed, no jargon that is not explained where it appears, and no claim made here that is not backed somewhere you can check. If you have five minutes and no particular agenda, start with the first one.
A day in the life of one organism
Born, hungry, briefly alive, gone. What actually happens to a single creature from the threshold crossing that created it to the corpse that returns its energy to the world, with a running note on exactly where the storytelling starts to lie.
How to watch a world
You clicked Watch, saw some slow dots, and left. Fair enough. This is the guide to the second visit: what to look at and in what order, which events are worth waiting for, and the viewing rhythm that makes a deliberately slow world worth the time.
Nothing in this world wants anything
Every simulation you have seen was trying to do something. This one has no score, no goal and nothing being maximised, and taking the objective away changes what you are looking at more than any other decision in the project.
What happens in one tick
The world does not flow, it advances in complete steps, and inside each one fifteen stages run in an order that never varies. Diagrammed in full, with the four things that ordering quietly decides, including why the dead never learn.
Where the energy goes
One currency, seven audited reservoirs, and a total that never moves. What it costs to exist, why failing costs full price, and why the universe halts rather than let the books drift.
Everything that has gone wrong
A world that died at tick 2042. Organisms apparently walking on the ocean. A staging server that ran the wrong world for months while everyone watched it. The failures, in detail, and what they have in common.
The science, beyond this experiment
These four stand on their own. They are about how evolution and ecology work in general, with the connection back to this world kept to a clearly marked section at the end of each.
How evolution actually works
Three requirements and no more: variation, inheritance, differential reproduction. The three modes of selection diagrammed, plus the four things almost everyone gets wrong, including the one about survival of the fittest.
Why populations boom and crash
Carrying capacity, overshoot, and the delay that turns a stable system into an oscillating one. Why the smooth curve in the textbook is the rarest case in practice, and when the recovery does not come at all.
Genetic drift, and why small populations lose variety
Chance alone fixes variants and erases others. Why drift can beat selection outright when numbers are low, what a bottleneck costs permanently, and why a headcount hides it.
How energy moves through an ecosystem
Matter cycles, energy does not. The ten per cent rule, and how one number explains short food chains, rare predators, and why the shape of life is a pyramid.
Evolution at larger scales
How one species becomes two
Speciation needs exactly one thing: inheritance stops flowing between two groups. The three routes drawn, why the process has no moment you can point at, and why biologists use several incompatible definitions of species on purpose.
Why the same solutions keep appearing
Eyes evolved dozens of times independently, flight at least four. The three constraints that narrow the space of workable answers, how you tell convergence from inheritance, and why it is not evidence of evolution aiming at anything.
How individuals became groups
Cells gave up reproducing alone and became bodies. Why selection should tear that apart, the three mechanisms that stop it, and the plain admission that no simulation, including this one, has produced such a transition.
How a neural network makes one decision
Multiply, add, squash. The whole arithmetic, followed by hand, why the squashing step is what makes depth mean anything, and the loop that gives a network a past to work from.
Life, inheritance and its limits
Why living things age and die
Not a failure of maintenance. Selection weakens with age, so late-acting damage goes unopposed, and some traits are actively favoured despite ruining you later. The theory makes a prediction, and bats confirm it.
Why sex exists, when it costs so much
Reproducing alone passes on twice as much and needs no partner. On the arithmetic sex should have lost long ago. The two-fold cost drawn, the two best answers, and why the question is still genuinely open.
How mutations actually work
The only source of new variation there is, and almost every instance is neutral or harmful. The types compared, and why gene duplication, the quiet one that breaks nothing, is where novelty actually comes from.
The hardest question in biology
Evolution needs inheritance, and inheritance needs something already copying itself. The circle at the centre of the origin of life, what the RNA world gets right, and why this project starts on the far side of the gap.
Ecology, evidence and how to read any of it
How living things stay stable
Being alive is an activity, not a condition: order held in place by continuous expenditure. Negative feedback, why tolerance is a band with sharp walls, and why a wide one costs more than it looks.
Living together: partners, parasites and rivals
Six ways two species can affect each other, why the categories leak so badly that they classify situations rather than species, and what actually stops a mutualist from taking the benefit and giving nothing back.
How we know evolution happened
Six independent lines of evidence that each had every chance to contradict the others and did not. Why the badly built traits, the blind spot and the giraffe's absurd nerve, are the most persuasive part of the case.
Why simulations mislead, and how to read one
The three ways simulated evidence goes wrong, a worked example from this project where every test passed and the picture was still false, and four questions worth asking of any result, including everything on this site.
Can an artificial ecosystem have a history?
Most simulated worlds have runs, not histories: start, watch, discard, repeat. What it takes for a world to have a past worth the word - one-way time, records that outlive what they describe, consequences that cannot be taken back - and the honest fact that most of such a history is nothing happening.
Why we refused to choose the most interesting universe
Three candidate worlds passed every test written down in advance. Taking the best of the three would have undone the whole method, so Universe 1 ran on the one that came first in a pre-registered order, with "interestingness" recorded as a non-reason.
The strait that closed
Universe 1 was founded as two islands with ocean between them. Its organisms have since moved 45,000 units of ground into that ocean and joined them, stripping the interior to the floor on the way. Nothing here rewards building, and the half of this that is real evolution is not the half you would guess.
Groups, cycles and scale
Coordination without a leader
A starling flock has no leader, no plan and no bird that knows the shape it is making. Three local rules, the correction real starlings forced on the model, and ants solving mazes with no ant knowing the route.
Why cooperation is hard
Four numbers trap two players who reason correctly in an outcome worse than they could have had. What changes when they meet again, why the winning strategy was the simplest one, and why biology runs into the same structure without reasoning at all.
Predator and prey: the cycle that never settles
Predators always lag behind prey, and the lag is the whole mechanism. Lotka and Volterra, the counterintuitive result about fishing less, and why the famous lynx data is less clean than the textbooks suggest.
Why big animals live slow
A mouse and an elephant each get roughly a billion heartbeats. The three-quarter power law, why the tidy surface-area story is not enough, and how one exponent sets the pace of an entire life.
These are essays. For the concepts on their own terms, see Concepts. For the implementation with real variable names and real numbers, see Explore. For the experimental record including the null results, see research notes.