Support
Background on the model, how to read the index, and what Fracture is — and isn't. If something here doesn't answer your question, write in.
Fracture is a single-screen iOS app that displays an illustrative composite index estimating the probability of a US civil conflict within the next four years. It is a discussion-grade educational tool that summarizes six structural factors drawn from the political-violence research literature into one readable number.
The headline percentage is the output of a weighted composite of six factors, mapped through a sigmoid function into a 0–100% scale. It is meant to be read as a relative indicator of structural pressure, not as a calibrated probability you could bet on. Think of it the way you'd read a fire-danger rating: a coarse signal about conditions, not a guarantee about outcomes.
The factor inputs are illustrative baseline values, not live data feeds. Fracture does not connect to the internet and does not pull from polling APIs, social-media firehoses, or news scrapers. The small day-to-day drift you see is a deterministic per-date perturbation around the baseline, designed to make the index feel alive without pretending to be a real-time tracker.
The six factors — Political Polarization, Institutional Distrust, Political Violence, Economic Stress, Partisan Animosity, and Extremist Activity — were selected because each appears repeatedly across the major academic literature on civil-conflict onset as a structural precursor. They are not exhaustive, but together they cover the dimensions most consistently cited by researchers who study how stable democracies destabilize.
The methodology screen inside the app cites three reference points: the Center for Systemic Peace civil-conflict threshold (a long-running coding framework for political-violence onset), Barbara F. Walter's How Civil Wars Start (2022), and Henrikas Bartusevičius (2014) on inequality and civil-conflict risk. These are starting points for further reading, not the entirety of the literature.
The headline number is bucketed into five bands: Stable below 10%, Elevated 10–18%, High 18–28%, Severe 28–38%, and Critical above 38%. These thresholds are anchored loosely to where the academic literature places meaningful inflection points; they are interpretive, not legal or operational categories.
Each day, Fracture applies a small deterministic perturbation to the baseline factor values so that the index varies slightly from one day to the next. The drift is bounded and reproducible — the same date will always produce the same number on every device — so the app feels current without claiming to be tracking live events.
The delta is the difference between today's composite and yesterday's, in percentage points. It is included to give a sense of short-term motion, but because the underlying drift is bounded and stylized, day-over-day moves should be read as texture rather than as breaking news.
The 30-day trend is a sparkline of the composite index over the past month of deterministic drift. It exists to provide visual context for today's number — to show whether the recent regime has been broadly flat, climbing, or falling — rather than to support extrapolation into the future.
The six factors are weighted: Political Polarization 20%, Institutional Distrust 15%, Political Violence 20%, Economic Stress 10%, Partisan Animosity 20%, Extremist Activity 15%. These weights reflect a judgment call about which dimensions the literature treats as most load-bearing for civil-conflict onset in consolidated democracies. They are illustrative defaults, not empirically optimized parameters.
Each factor is normalized to a 0–1 scale, multiplied by its weight, and summed into a composite score. That score is then passed through a sigmoid (logistic) function to produce a value between 0 and 100%. The sigmoid compresses extreme inputs and centers the working range, which is why the baseline tends to land in the upper-High band rather than at either end of the scale.
It doesn't, directly. Researchers like Barbara Walter, the Center for Systemic Peace, and Bartusevičius use methods that range from structured expert coding to large-N statistical models on cross-country datasets — far more rigorous than what an illustrative single-screen app can produce. Fracture's contribution is presentational: it puts the kinds of factors those researchers care about in front of a general reader in a single, legible view.
No. Fracture is an illustrative composite, not a forecast. It does not output a calibrated probability, it is not validated against historical conflict-onset data, and it is not designed to be acted on.
No. Fracture is intended to spark discussion and to help readers think structurally about the factors researchers associate with civil-conflict risk. It is not a basis for financial, relocation, security, or political decisions.
Rigorously, only with caveats. The academic literature treats civil-conflict onset as a rare-event problem with limited historical data per country, which makes precise probability estimates inherently uncertain. Fracture sidesteps this by framing its output as illustrative rather than predictive — the percentage is a way to summarize structural pressure, not a claim about an underlying physical probability.
The app is non-partisan by design. The six factors are framed as structural characteristics of a polity rather than as positions held by any specific party or movement, and the methodology screen reads the same regardless of which side of the aisle you sit on. The point is to summarize academic concepts, not to advocate.
No. Fracture is fully offline. It makes zero network requests, ever. Everything you see on screen is computed locally from the bundled factor definitions.
No. Fracture collects nothing, stores nothing about you, and contains no analytics, crash reporters, or third-party SDKs. There is also nothing for you to enter — Fracture is a pure-display app. See the Privacy Policy for detail.
No. Fracture v1.0 is a free, one-time download with no in-app purchases. If a future major version ever adds paid functionality, it will be a one-time unlock, not a subscription, and the version you already have will keep working as-is.
Email OmarAppSupport@gmail.com. For bugs, include your iPhone or iPad model, iOS version, and Fracture version. For feedback on the methodology or the factor choices, just send your thinking — pointers to additional literature are welcome.