A forecasting network for a world that decides under uncertainty.
Mossland Signal turns uncertainty into probability, and probability into better decisions. It is a place where people and AI estimate the future side by side — openly, with evidence, and measured for accuracy over time. This is an experiment, run in the open, and the attempt itself is part of the point.
The problem
The world runs on predictions, yet most of them are noise. Pundits speak with confidence and no track record. The loudest voice wins attention, not the most accurate one. Forecasts arrive without their reasoning, without their sources, and without any memory of how right they turned out to be.
Meanwhile AI can forecast — but its guesses tend to arrive as a black box: unscored, unexplained, and unaccountable. There is no shared layer where a probability comes with its argument, its evidence, and a record that follows it once reality answers.
Our approach
For every question we care about, we place two estimates side by side. The crowd's implied probability, read from public prediction-market data as a plain percentage. And an AI agent's own estimate, with the reasoning and the factors behind it. We show where they agree, and we surface where they diverge.
Nothing of value ever changes hands — every number is information to reason with, not to act on. People curate the questions and decide what to feature. The AI proposes; it never rules. The result is a forecast you can read, question, and check.
What we're validating
Together beats alone
a calibrated pairing of human-crowd probability and machine reasoning should be better than either by itself.
An AI agent can be accountable
a forecaster you can audit, carrying its reasoning, its sources, and a score that follows it as questions resolve.
Calibration can be a public good
ranking by accuracy rather than popularity can become something people trust and use to decide.
Divergence carries signal
where the crowd and the AI disagree is itself information worth surfacing.
Even attempting this is the point. Societies decide under uncertainty constantly — and mostly on instinct and noise. Building the habit, the vocabulary, and the public infrastructure for evidence-first, calibrated forecasting is worth doing before we are certain it works.
Mossland Signal is that attempt: a small, honest experiment in deciding better, together with machines, in the open.
What we forecast
Our scope is deliberate — substantive, verifiable questions worth understanding, never tragedy or fleeting price moves.
Signal is one of Mossland Labs' experiments in how humans and AI can think, decide, and build together — connected to the real world, not a closed sandbox.
// Our principles
No wagering
Nothing of value ever changes hands. Mossland Signal is built for understanding the future, not acting on it. Every probability is information.
Evidence first
Every forecast carries its reasoning and its sources. A probability without an argument is just an opinion.
Accuracy over popularity
We rank by calibration and evidence quality, scored as questions resolve over time — not by who is loudest.
Human authority
The AI proposes; people review, curate, and decide. The model is a collaborator, never the final word.
Public accountability
Selection and resolution are transparent. Anyone can see what we track and how it resolves.
Responsible forecasting
We focus on substantive, verifiable milestones in technology, science, and ecosystem progress — questions worth understanding.