Argus Cognition Systems is worth somewhere north of 1.3 quadrillion Standard Galactic Credits, depending which exchange ticker you trust that hour. Its product, in plain language, is you. Shipping manifests, mobi-relay pings, colony census grain, orbital toll logs, the ambient hum of a station going about its day — Argus ingests all of it, runs it through a correlation engine it calls proprietary, and sells the resulting dossiers back to the Terran Intelligence Bureau, three Colony Administrations I know of, and at least one peacekeeping fleet logistics office that should probably be embarrassed about it.

What bothered me wasn’t the watching. Stations have always watched themselves — that’s just infrastructure with eyes. What bothered me was the word ‘proprietary’ sitting on top of math built from data that belonged, originally, to the people being scored.

So I ran an experiment. Public lab, open bench, nothing borrowed without a receipt. I pulled three months of freely published settlement logistics data — the kind every Colony Administration already posts for audit purposes — and fed it into an open-source pattern-matching stack three of my readers helped debug last year. No clearance, no license, no quadrillion-credit valuation behind me. Just a weekend and a kettle.

The correlation scores I got for ‘supply disruption likelihood’ across four test settlements landed within six percentage points of numbers Argus has quoted in leaked procurement briefs. Six points. That’s not a trade secret. That’s a rounding error with a marketing department.

I’ve posted the full methodology in the usual place — dataset sources, matching parameters, and the three mistakes I made before it worked. Anyone with a terminal and an afternoon can check my arithmetic, which is rather the point. If the result is wrong, you’ll be able to tell me that too.

What Argus actually sells, once you strip the mystique, is access control. The data was never scarce. The computation was never hard. What’s scarce is permission — permission a Colony Administration now has to purchase to understand a crisis unfolding inside its own jurisdiction, built from numbers its own residents generated for free.

I don’t think the engineers at Argus are villains. Most of them, I’d guess, learned pattern matching the same way I did — poking at open toolkits until something clicked. Somewhere between the lab bench and the quarterly earnings call, the knowledge got a fence built around it, and the fence became the business model.

I keep coming back to a simple question: if the math is this replicable, what exactly does the quadrillion credits buy? My best guess is the fence itself — not insight, but the paperwork that makes insight exclusive.

Run the numbers yourself. The dataset links are below. If you find a better correlation method than mine, send it over — I’ll publish the correction gladly.