Structure, not personality
The engine models capacity, trade, prices, energy, demography and alignment. It does not model what one leader decides on one afternoon, and it never claims to.
What it is not
Not a forecast, not a dashboard, not a spreadsheet
The clearest way to say what a tool is, is to say what it refuses to be. Here is where Futurisca sits next to the three things people reach for instead.
Side by side
| A forecast | A dashboard | A spreadsheet | Futurisca | |
|---|---|---|---|---|
| Answers with | One number | What already happened | One number per set of inputs | A distribution over hundreds of futures |
| Shows its reasoning | No | No | Formulas, if you dig | An exact decomposition into named terms |
| States uncertainty | A confidence interval, sometimes | No | No | Branches with a share each, plus sensitivity |
| Handles second order effects | No | No | Only what you wired by hand | Yes |
| Countries covered | A handful | Whatever you loaded | Whatever you typed | 238, none skipped, down to the county and the town |
| Lets you change the world | No | No | Yes | One sentence, no template |
| Steps through what follows | No | No | No | Year, month, week or day, with the reasoning at each |
| Says what to change | No | No | No | Twelve candidates, each re-run against your scenario |
| Comparable runs | No | No | Only if you are careful | Same seed, same shocks, difference is your lever |
| Says where a number came from | Rarely | Sometimes | No | A provenance mark on every row |
It is not a crystal ball. It is a computational laboratory for the future.
Where the boundary sits
A model, not a measurement
The starting world is as accurate as the sources allow and is marked accordingly. Everything after year zero is the output of a model, and the interface says so on every screen.
Ranges, not points
Where a number is quoted it comes with the spread around it. If you want a single figure to put in a headline, this is the wrong instrument and it will keep telling you so.
Four things it gets mistaken for
Every one of these is a reasonable thing to want, and none of them is what this is. Knowing which one you actually need is worth more than any feature comparison.
01
An AI that predicts the future
There is no model making a prediction anywhere in this. The engine is arithmetic: published elasticities, a trade network, drawn shocks and a Monte Carlo over them. If you want a system that reads the news and tells you what happens next, this refuses to be it, and it refuses on purpose.
02
A live operations picture
The map carries live feeds and they are genuinely live, but they sit beside the model rather than inside it. Nothing on a feed changes a run. If your question is what is happening right now in a place you are responsible for, you want a monitoring product, and this will not replace one.
03
An economic research platform
The coefficients are published values with published ranges, not estimates produced here. The model is stylised on purpose: annual steps, no real financial sector, trade as a network rather than a ledger. It is built to find the shape of an effect, not to publish one.
04
A decision that has been made for you
It returns a distribution, a set of families, a decomposition and a confidence score. None of those is a recommendation. The instrument is deliberately silent about what you ought to do, because the point of showing the arithmetic is that the judgement stays yours.
When to use something else
You need a number for a board slide by Thursday
Use a published forecast from an institution whose name carries the authority you need, and cite it. This will give you a range and a set of caveats, which is more honest and less useful for that specific purpose. It is worth being clear about that.
You need to know what happened last quarter
Use a statistics agency. This is a model of futures conditional on changes that have not happened. Its base year is as good as open data allows and it is not a reporting tool.
You need firm level or product level supply chain detail
Use a trade data product. Trade here is a bilateral network over a small set of goods categories, which gets the direction and rough size of a shock and cannot tell you which supplier in which province is exposed.
You need something defensible in a regulatory filing
Use a model with an audit trail, a validation history and a named methodology owner. This publishes its assumptions in full, which is not the same thing, and it says so rather than implying otherwise.
You need to understand the shape of a decision
This is what it is for. Which channels a policy acts through, who else it reaches, how much of the answer is a coefficient nobody knows, what would have to be true for the answer to flip, and which of the futures it produces you would actually be planning for.
Read the method before quoting a number
Every assumption, prior and known weakness is written down.
Method and limits