Signals without language
Every organism can broadcast one number, and its neighbours can sense it. That is the whole mechanism. Whether any of it amounts to communication is a question this project answers by measurement rather than by looking at the pictures - and the most useful thing measurement has produced so far is a demonstration that the measurement itself is not strict enough.
What an organism can actually transmit
One continuous, signed value between -1 and 1, emitted through a single action channel, once per tick. Not a word, not a symbol, not a choice from a vocabulary: a magnitude and a sign. An organism has no way to transmit anything else, and no way to direct it at a particular recipient.
It is not free. The emission costs energy in proportion to its magnitude, which matters more than it sounds: a channel that costs nothing can drift into use for no reason, and one that costs something has to earn its keep against everything else that energy could have done. Signalling in this world is always paid for by the sender.
How a neighbour receives it
Through two sensory channels, both generic, neither labelled as "communication":
- The signal field - what was broadcast at and around this location on the last tick, ranging from -1 to 1. It is instantaneous and it decays away.
- The trace field - a persistent, decaying deposit left behind by past signalling in a place, ranging from 0 to 1. This is the closest thing the world has to a mark left in the environment rather than in the air.
A receiving organism sees those two numbers the same way it sees light, elevation or resource density: as inputs to the same network, with no privileged status. Nothing tells it that a signal came from an organism, or which one, or that it means anything at all. See the full channel list for where these sit among everything else an organism senses.
Why this is not language, and what would have to be true first
A signal carries no built-in meaning. Nothing in the rules associates any value with food, danger, kinship or location. If a value ever comes to mean something, that meaning has to be built by evolution on both ends at once - a sender whose emission depends on something real, and a receiver whose behaviour depends on what it heard. Either half alone does nothing.
That is a measurable claim rather than a philosophical one, and this project has written the threshold down rather than leaving it to judgement. Two conditions, both measured server-side against the live world, continuously:
- senders carry at least 0.05 bits of excess information in their emissions - that is, the signal says something about the sender's situation that could not be guessed without it;
- and at least 5% of receivers measurably respond to what they heard.
Both must hold together, because either alone is nothing: a sender broadcasting real information nobody acts on is talking to itself, and a receiver reacting to noise is superstition.
The glyphs on Watch are gated on exactly this measurement, recomputed continuously against the live world. So the honest live answer is visible rather than asserted: if you can see glyphs over organisms at close zoom, the world is currently passing. As of late September 2026 it is - comfortably, with the signal carrying most information about the sender's own health. A member can also switch the glyphs on as a research view when the world is not passing, which is why seeing one in a Lab session proves less than seeing one on the public page.
Why the glyphs are a rendering vocabulary, not an alphabet
At close zoom, Watch can draw a small glyph above an organism whose emission has just changed. It is easy to read more into these than they carry, so it is worth being exact about what they are.
The continuous emission is bucketed into five fixed bands for display. Each band has one glyph. That mapping is identical for every organism and every lineage in the world, and it was chosen by the people who wrote the renderer, not by anything in the simulation. A glyph therefore always means one thing - "this organism is currently emitting in this band" - and never anything about intent. The middle band is silence and draws nothing.
| Glyph | Emission range | What it tells you |
|---|---|---|
| ∴ | -1.0 to -0.6 | Strong negative emission |
| ◦ | -0.6 to -0.2 | Weak negative emission |
| (nothing) | -0.2 to 0.2 | Silence. No glyph is drawn |
| ∿ | 0.2 to 0.6 | Weak positive emission |
| △ | 0.6 to 1.0 | Strong positive emission |
No lineage has its own glyphs, and none can acquire them. If something like a dialect ever appeared, it could only show up as different patterns of use of the same shared symbols, never as a private alphabet. The glyphs are a way of making a hidden number visible to you. They are not a writing system, and the organisms cannot see them.
What the evidence currently says
A registered study asked directly whether signalling pays, and answered with a control most informal claims never face: the same worlds run three ways, with signals heard normally, shuffled between organisms every tick so they became noise, and silenced entirely.
Receivers in the normal worlds did not respond to their neighbours - at most 0.19% in any measurement window. The worlds where signals were audible did not outperform the controls; where all three finished, they had the smallest populations. Lineages did not develop measurably different uses of the bands.
The most useful result was an embarrassing one. One shuffled control - a world where signals had been deliberately randomised into meaninglessness - passed this site's own signalling test in its final window. That is not evidence of communication in a scrambled world; it is evidence that the test was not strict enough, and it is recorded as such rather than quietly dropped.
Read the full study, including its limitations - chief among them that roughly 120 generations is short, and nothing in it rules out signalling appearing over a far longer run.
So why are the glyphs on?
Because the live world passes the criterion, and has done comfortably. That sits awkwardly beside the study above, and the awkwardness is the most honest thing on this page.
Universe 1 ran far longer than any of those runs - hundreds of thousands of ticks rather than tens of thousands - so it is not the same measurement, and a longer world genuinely might do something a shorter one did not. That is the optimistic reading and it cannot be ruled out.
The unglamorous reading is stronger, and it comes from this project's own data: a shuffled control passed this same test. A world where signals had been deliberately scrambled into noise cleared the bar the live world is now clearing. Whatever that bar detects, it is not necessarily communication, and a passing score is therefore necessary but nowhere near sufficient.
So the correct reading of a glyph on screen is narrow: an organism is emitting in a band, and the world currently clears a threshold known to be clearable by noise. It is not a word, it is not evidence that anything was understood, and nobody here is claiming otherwise.
Passing the criterion is not evidence of meaning, because a randomised control passed it too. Nothing here shows that meaningful signalling has emerged in Universe 1, and nothing shows it is impossible either. The honest position is that the current test is too weak to separate the two, that this is known rather than hidden, and that a stricter test is the work - not a better picture of a glyph.
See it for yourself
Watch the live world, select any organism, and its signal emission is visible in the Brain Inspector alongside every other channel it senses and acts on - free, no account needed. How that panel works · every channel an organism has · what would count as emergence here