themildtake

an engineer's analysis

The systems that move money, power, and risk.

The Mild Take publishes source-disciplined analysis: long-form pieces on AI infrastructure, technology, and markets, plus a confidence-weighted country-risk ratings tool for where to live, hold assets, and hold currency. Multilateral and independent sources triangulated, US press treated under explicit ownership-tilt flags, ranges stated honestly.

Read the latest → Explore the ratings

latest

Recent writing

All articles →
Analysis

Refresh: a courtroom, a tariff, and a ceasefire on the table

A one-day window, so most of the board held. But the day carried three things worth recording: a US federal judge ordered the National Guard out of Washington, reversing the deployment we had scored as a fresh institutional low; Washington hit Canada with 50% tariffs; and, even as the tenth straight night of strikes fell on Iran, a 10-day ceasefire proposal reached Tehran — the exact exit the framework refused to score away. Seventeen countries moved.

  • live-data
  • refresh
  • iran
Analysis

Refresh: the war that didn't break

Last pass caught the ceasefire cycle snapping into escalation. This one records the escalation refusing to break: a ninth consecutive night of US strikes on Iran, the first American combat deaths of the war, Iran hitting Kuwait's water and power for the first time, and the Houthis opening a maritime blockade on Saudi Arabia. Around it, the EU froze the Russia oil cap rather than let it lapse, Nicaragua's Ortega declared there would be no more elections, DR Congo's Ebola outbreak began outrunning its own contact-tracing, and a claim we excluded on sourcing turned out to be true. Thirty-four countries moved.

  • live-data
  • refresh
  • hormuz
Articles

Unstable Legion: the peer-to-peer ChatGPT

Unstable Legion is a peer-to-peer ChatGPT: one language model cut into layer-stages that run in ordinary web browsers and hand work to each other over the Codec wire, with no data center in the loop. Open a tab, chat, and optionally donate your GPU to host a few layers. The pipeline assembles itself across whoever is online and keeps generating as peers come and go. It is early and experimental, and it runs today at legion.codecai.net. Here is how it works, what is proven, and where it is headed.

  • codec
  • legion
  • mesh-inference

the ratings tool · living decision

Country risk for where you live, invest, and hold money

One of the things we publish is a confidence-weighted country-risk ratings tool: a directional read from −10 to +10 with explicit confidence, for 194 countries across the living, asset, and currency decisions. Open any country for the full breakdown, or compare them side by side.

  • +3 to +10 (strong)
  • +1 to +3
  • -1 to +1 (neutral)
  • -3 to -1
  • -10 to -3 (weak)
Hover any country for rank, score, and confidence. Greens are stronger institutional + practical reads for relocation; reds are weaker; yellow is near-neutral. Open the full ratings app for the per-decision grid and category breakdowns.

Directional, not gospel. These are confidence-weighted reads calibrated to one decision-maker's priorities — the methodology explains how each score is bounded. The asset and currency reads are analysis, not financial advice.

Articles →

Distilled, data-first pieces on AI infrastructure, technology, and the structures moving money and power — one argument each, with the numbers that carry it.

Analysis →

Long-form, source-disciplined investigations — the full evidence, the ranges, and where the skew points.

The ratings app →

A live grid of every country, per-decision rankings, category radar charts, and country-vs-country comparison.

Resources →

The source canon behind the work — bias and coverage checking, integrity data, open datasets, fact-checking — with a live snapshot of what's changed.