How this map is built
ForesightGraph follows how a political or economic development reaches people, and it keeps the evidence for every step in view. This page says where the content comes from, who checks it, and what the map does not tell you.
What a line means
Every line is a claim that somebody wrote down, tied to sources and put up for review. A person had to approve it before it appeared here. It is not a similarity that a program found between two sets of figures.
That also sets the limit. A line says that someone claims a direction of impact and names what carries it. It does not say that the impact has been measured, and whether other causes contribute usually stays open. The size of an effect is known only where a source states it.
What you are looking at, and what you are not
The map shows one published version, and drafts are not on it. Its nodes and connections were copied when it was published and keep the wording and the ratings they had on the date shown beside the version. The evidence behind them was not copied: sources, quotations and claims are read as they stand today, so a later correction to one of them also changes what this version shows. New findings go into a new version.
This graph is deliberately small. It holds what could be checked against a source, and nothing beyond that. A missing connection means that it has not been checked, not that it does not exist.
What is covered is a corridor and not a whole field: one region, a few base systems, a few user roles. Whether a topic is covered completely is a question this portal cannot answer about itself.
If the header says Demo dataset
A data set can be marked as a demo, and the marking stays in the header for as long as it is on screen. Its sources are invented and its quotations were written for the demonstration. Nothing in it is a statement about the world. What can be checked on it is the method: the chain is complete, and it is the same chain that content from real sources goes through.
Who writes this, and who decides
Software drafts. People decide. A draft becomes visible here only after a person has approved it, and no automated step publishes anything or deletes anything for good.
Approval and publication lie with the operator of this portal. There is no independent panel behind them.
The rule holds where it costs something. A claim without evidence can be approved only with a recorded reason for the exception, and a relationship that rests on such a claim cannot be published. Such cases occur, and they hold a publication up until somebody resolves them.
What sits behind a line
Seven steps lie between a document and a line on this map. They are kept apart on purpose, because each one can fail in its own way.
- Source.
- Who published something, with publisher and date.
- Snapshot.
- The document as it was on the day it was fetched, stored with a checksum. A later change to the original cannot quietly alter it.
- Evidence.
- The passage, quoted word for word and in the language of the original. It is not translated, because a translation already carries a reading.
- Claim.
- The sentence the portal asserts, with its type, its confidence, the region and the period it covers.
- Relationship.
- The assertion that one thing acts on another, together with the claims that carry it.
- Impact path.
- Several relationships in a row, ending at a group of people.
- User impact.
- What arrives at people, for a stated region and a stated period.
The break is rarely at the beginning. A chain can be sound at the source, the snapshot and the quotation and still have a hole at the fourth step. Reading the name of a publisher and taking the chain for closed is the mistake this layout is built against.
A chain holds only as far as its weakest step, and the portal does not work that out for you. There is no single figure for how far a path carries.
Four kinds of claim, four questions
Every claim carries one of four types. The type is not a filing category. It says which question is the right one to ask.
- Fact.
- A measured or officially recorded figure. Does it cover the same region and the same period as the line it supports?
- Forecast.
- A statement about a period that has not been measured yet. How far ahead does it reach, and what does it assume?
- Assumption.
- A step nobody has evidenced and the argument needs anyway. Do you accept it? Nothing else holds it up.
- Interpretation.
- A conclusion drawn from several figures that the source itself does not draw. Does the way from the figures to the conclusion hold?
In the detail view the same word also labels the note an analyst adds beside a quotation. That note is not a claim of its own.
What the ratings say
A relationship can carry four ratings: strength, speed, duration and confidence. Strength and confidence are always set, speed and duration may be left open. A level for speed or duration cannot be saved without a recorded reason for it. For strength and confidence a reason stands beside the level where somebody wrote one, and nothing forces it.
Where a level is missing, the detail view says “Not rated” and keeps the row instead of dropping it. That is not the bottom of the scale. It says that nobody has determined the level, and for speed and duration that is the usual case rather than the exception.
The four are never combined into one score. The weights for such a score would be ours, and no source would carry them.
Where the qualifications are, and how to reach them
Uncertainty is not tidied away. Confidence, region, period, how recent the sources are and the limitations of a claim all sit at the object itself. Evidence that contradicts a claim stands next to it instead of being removed. Select a node or a connection to read it, on the map or in the list view.
One thing you will not find is a figure for the size of an effect where no source gives one. Where that is the case, it is stated in the limitations.