GBTT Wiki

Three pillars support the thesis. State capacity failure: the state is bigger than ever and delivers less, with failures simultaneous across unrelated sectors (justice, borders, health, housing, infrastructure). Social reproduction failure: fertility at 1.41 TFR, home ownership for young people collapsed, productivity stagnant since 2008. Democratic legitimacy failure: 25 years of 60-70% opposition to high immigration, net migration rose from 50k to 700k across four changes of government.

Key concepts: the deontological shield (correct procedure producing monstrous outcomes with no accountability), bifurcation (global-facing domains thrive while domestic-facing domains decline), and the 1970s comparison (that crisis produced a democratic correction; this one hasn't, because legitimacy crises self-reinforce rather than self-correct).

Falsifiable by 2033: if 3+ of these improve significantly, the thesis is wrong. Trust in government above 25%. Fertility above 1.6. Under-35 home ownership above 45%. NHS waiting list below 4m. Net migration below 200k for 3+ years. Crown Court backlog below 40k.

The card is the product. A single image, one killer stat, designed to drop into an existing X or Reddit argument and win it. The funnel: argument > card > tool/article > GBTT brand. Cards are extracted from GBTT's novel research, not generic headlines. Each card flips the moral framing: data IS compassion when you're showing who actually gets hurt.

The framing pattern: place the WARNING label on the policy that's actually hurting people. Reframe the data-driven position as the compassionate one. "Compassion" says protect pensioners; the card asks who needs protection more, a triple-locked homeowner or a 28-year-old paying £1,430/month rent.

Every piece goes through a 10-phase pipeline before publication. The pipeline is designed to prevent the single most damaging thing a data-driven outlet can do: publish a wrong number.

For articles (not tools), an additional quality pipeline runs after phase 9:

Confidence tiers: Strong Signal (3+ sources or primary source confirmed), Moderate Signal (2 sources), Speculative (1 source). Target: 80%+ Strong Signal before publication. No unverified claims without explicit Speculative labels.

Karpathy autoresearch loop: frozen metric (Strong Signal / total claims x 100), git as memory, keep/discard binary, max 5 iterations. The loop attempts to upgrade every Moderate and Speculative claim.

Every GBTT topic connects to the elite decoupling framework. The reader who sees one post gets value. The reader who sees ten starts to see the pattern.