THE EVIDENCE.
Every day, people make choices that change what happens next. We report on those choices, check the facts, and ask one simple question: What future are these decisions making more likely?
What does a score actually mean?
Think of the percentage as our best estimate of how likely a particular future event is to happen by a stated date. A score of 90% means we think that outcome is very likely. It does not mean the story is 90% true, that an action is 90% good or bad, or that we can predict the future with certainty.
For example: 90% GOOD might mean we estimate a 90% chance that a named clean-energy target will be reached by December 2027. 90% BAD might mean we estimate a 90% chance that a specific pollution target will be missed by that same date. Both scores describe future outcomes, not opinions about whether the news itself happened.
Another way to picture 90%: if we made 100 comparable predictions rated 90%, a well-tested system would expect about 90 to happen. Our scores are editorial judgments, not yet scientifically tested or proven to meet that standard. We can be wrong.
Why are stories only shown at 85% or higher?
We want the front page to focus on consequential outcomes we currently believe are very likely, rather than every alarming headline or hopeful possibility. Each side has the same minimum score of 85%. A high score does not make a story more important than another, and it is not proof the prediction will come true.
Where does the evidence come from?
We start with reporting from around the world, then check important claims against original documents, scientific research, public data, and other reliable sources wherever possible. Each article links to the material behind it. We separate what has happened from what we think could happen next, and explain what could make our prediction wrong.
Can an article or its score change?
Yes. The score is a living estimate, not a permanent verdict. New evidence can change the forecast in either direction. A government might reverse a decision. New research might challenge an assumption. Fresh data might show a policy working better—or worse—than expected.
When important new information arrives, our editorial process is to recheck the sources, review the original prediction, and update the article and probability when warranted. An article can move from 91% to 87%, rise from 86% to 94%, or fall below our 85% front-page threshold. If it falls below 85%, it stops appearing on the Good or Bad front page; the original article link can still be visited.
Articles and assessments now live in our version-controlled codebase. Editors review sources and commit corrections so changes can be audited in Git. We are building toward showing a visible history of those changes beside articles. The public change-history display is not live yet.
What happens when the forecast's deadline passes?
We should check the result against the exact claim and deadline, and mark it as happened, did not happen, or not yet clear. As the publication grows, we plan to compare our past scores with actual outcomes and publish how accurate our forecasting has been. Until that record exists, please treat every percentage as an informed estimate—not a promise.
What about the first ten articles?
They are sourced, production-style launch examples included with the website. Their numbers illustrate our process but have not been independently audited or statistically calibrated. Sources are credited for their reporting and evidence; they do not endorse our articles or scores.
See an error, missing source, or a new development? We believe evidence should change minds. Reviewing and correcting our own work is part of the publication's purpose.