Everything runs in your browser. Nothing you build is sent anywhere.
The simulator ships with a state of every country, and for most fields it is as good as open data gets. It is still somebody else's data. This page says exactly what to upload, where it goes in, what happens to each row, how a figure of yours is marked, and how it reaches the model.
An upload replaces base year figures for the countries and fields you name. It does not add a field, change an equation or touch a lever. Every run after it, baseline and scenario alike, starts from your figure instead of the one that shipped, and the arithmetic runs on from there exactly as before. That is the point: the reasoning is unchanged and the input is yours.
Everything you put in is checked before it is stored, marked as yours wherever it appears, and can be removed in one press.
In the WordPress admin, under Futurisca, the screen is called Your data. It is for administrators: the capability it checks is the one that manages options. The screen shows how many figures from you are live in the world now, how many fields may be set, and how many countries the model carries.
Download a starting file hands you a CSV in the exact shape the reader wants back, filled with the figures the world holds now for output, population, inflation and unemployment in three economies. Change the values that are wrong, add the rows you have, delete the rest. An empty template is a worse start than a filled one.
A file chooser, or the same text pasted into the box beneath it. Whichever is filled in is used, and the file wins if both are.
The first button reads the whole file and reports what would happen without writing anything: every accepted row with the value the world holds now beside the one you are giving it and the percentage between them, which is the fastest way to catch a decimal point in the wrong place, and every refused row with its line number and the reason. Two hundred corrections should be looked at before they are in the world.
The second button runs the same check and then stores every row that passed. Figures already uploaded for the same country and field are replaced. Nothing is written until the whole file has been read. The report stays on the screen for ten minutes.
Remove everything I have uploaded clears every figure that came from an upload and nothing else. Values written by a World Bank refresh are left alone.
A CSV, one row per figure: a country and a field. The header row names the columns, in any order, and reads iso, field, value, year, note. Commas or semicolons both work, a byte order mark is stripped, and thousands separators inside a number are handled, so a file exported straight out of a spreadsheet works without being cleaned up first.
The same country and field twice in one file is refused on the second line rather than letting the last one quietly win.
The range beside each field is a refusal boundary rather than a style guide. A value outside it is nearly always a unit mistake, and a unit mistake reaching the model produces a confident wrong answer, which is the most expensive thing this instrument can do. The list is read from the simulator itself, so it cannot drift from what the upload accepts.
| field | Reading | Unit | Admitted range | Note |
|---|---|---|---|---|
| pop | Population | millions of people | 0.0001 to 2000 | People, in millions. A country of 8.4 million is 8.4, not 8400000. |
| gdp | Output | USD billions, nominal | 0.01 to 60000 | Nominal GDP in billions of US dollars. 30438 is thirty and a half trillion. |
| gdppc | Income per head | USD | 100 to 300000 | Nominal GDP per person, in whole dollars. |
| infl | Inflation | percent a year | -30 to 400 | Consumer prices, percent. 2.7 rather than 0.027. |
| unemp | Unemployment | percent of the labour force | 0 to 70 | |
| rate | Policy rate | percent | -5 to 200 | |
| debt | Public debt | percent of GDP | 0 to 400 | |
| expgdp | Exports | percent of GDP | 0 to 250 | |
| agri | Agriculture share | percent of GDP | 0 to 80 | |
| ind | Industry share | percent of GDP | 0 to 90 | |
| serv | Services share | percent of GDP | 0 to 95 | |
| urban | Urbanisation | percent living in cities | 0 to 100 | |
| popg | Population growth | percent a year | -8 to 12 | |
| age | Median age | years | 10 to 60 | |
| life | Life expectancy | years | 30 to 95 | |
| fert | Births per woman | children | 0.3 to 9 | |
| internet | Connectivity | percent online | 0 to 100 | |
| rnd | Research spend | percent of GDP | 0 to 12 | |
| mil | Military spend | percent of GDP | 0 to 40 | |
| ai | AI capability | index, 100 is the leader | 0 to 150 | |
| semis | Fab capacity | percent of world | 0 to 100 | |
| encons | Energy use | exajoules a year | 0 to 250 | |
| enprod | Energy produced | exajoules a year | 0 to 250 | |
| oil | Oil production | million barrels a day | 0 to 30 | |
| renew | Renewable share | percent of energy | 0 to 100 | |
| co2 | Emissions per head | tonnes a year | 0 to 80 | |
| water | Water stress | index, 0 to 100 | 0 to 100 | |
| food | Food self sufficiency | ratio, 1 is self sufficient | 0 to 4 | |
| stab | Stability | index, 0 to 100 | 0 to 100 | |
| gov | Governance | index, 0 to 2 | 0 to 2 |
With its line number and the reason: a code the model does not carry, a field that cannot be set, a value that is not a number, a value outside the admitted range, a year outside 1960 to next year, or the same country and field a second time. A file of two hundred rows with three faults stores a hundred and ninety seven.
A file that goes wrong halfway through cannot leave the world half corrected. On a check, nothing is stored at all.
The header row needs iso, field and value. Missing one, the report says which and lists the columns it found. A file with no rows under its header is refused the same way.
The country record carries, beside the value, a source mark reading upload and the year you gave, and a list of the fields that are yours. The mark is part of the record printed into the page, so wherever the record goes the app can tell your number from the World Bank's.
On the Monitor tab, the section Where these numbers come from counts the countries carrying figures you uploaded, and its country by country table has a column, Your own fields, that names them.
Your data keeps a count of the figures from you that are live in the world now, and the last report, accepted and refused, stays on the screen for ten minutes after it was run.
Uploaded figures are merged into the country records the moment they are read, and those records are the base year state the app is handed. Every run compiles its world from that state, so every run after the upload starts from your figure: the baseline, the scenario, the comparison, all of it.
Output divided by population, at load and again inside the engine. An uploaded income per head is accepted and stored but recomputed from the other two, so to move it, supply output and population.
Births per woman, life expectancy and urbanisation are carried on the record and shown on the map and in the monitor, and the engine does not step them. Population growth, median age and population are what the demographic block runs on.
The refresh writes the seventeen fields the World Bank publishes and replaces yours for those: population, output, inflation, unemployment, exports, urbanisation, population growth, research spend, military spend, the three sector shares, life expectancy, births per woman, connectivity, emissions per head and the renewable share of electricity. The other thirteen survive it. Uploading again puts yours back.
No coefficient, equation or lever moves. What changes is the world the equations start from, and the marking says so.
Which base year figures a question rests on, read from what the engine actually uses. Supplying a field the question does not rest on is harmless. Leaving one out means that part of the answer runs on the figure that shipped.
| The question | The fields to supply | Why those |
|---|---|---|
| What a tariff, a control or a sanction does here | gdp, expgdp | The trade network is built from output over distance, and the share of output that leaves sets how much of the economy a trade shock can reach. |
| Where prices, rates and jobs go | infl, rate, unemp | The Phillips curve, the Taylor rule and Okun's law each start from the figure on the record. |
| How the debt path and the risk premium look | debt, stab, gdp | Debt dynamics run on the stock against output, and the premium starts from the debt and the stability on the record. |
| What an energy shock costs | enprod, encons, oil, renew | Dependence is produced against used, oil supply answers the price, and the renewable share sets how much of the economy the fossil price reaches. |
| What a chip shock does | semis, ind | Fab capacity says who makes them. The industry share says who cannot do without them. |
| How fast capability arrives | ai, internet, rnd | Capability spreads at the speed a country can absorb it, and connectivity and research spend set that speed. |
| Who is under strain, and who moves | stab, water, food, agri | Migration pressure reads stability, water stress and food. The agriculture share says how hard a food price lands. |
| What arrives in twenty years | pop, popg, age | The labour force is population against median age, and population growth is the base year rate. |
| What a carbon price costs, and to whom | co2, renew, gdp | Emissions per head against output is the bill, and the renewable share is what escapes it. |
| What a defence build does to the budget | mil, debt | Military spend is paid for out of the primary balance, which is where it meets the debt. |
The camera layer draws a curated set of forty public cameras that needs no key, and the settings add to it. On the Futurisca settings screen the Live cameras field takes one camera a line, in six parts separated by a vertical bar: name, latitude, longitude, an https image URL, the refresh in seconds, and a credit. The URL must be https and must return an image, which is what almost every public authority's traffic, harbour and weather camera already does. A line starting with a hash is ignored.
Optional beneath it: a Windy Webcams key. With one, the server fetches a page of public cameras on each refresh and caches it, so the key never reaches a browser and no visitor ever talks to Windy.
The six parts are separated by a vertical bar, in that order.
What is held for every country and every place, and where each figure came from.
The world data