Friday, January 16, 2026 · 124 newsletters

ICE Started the Fire

politics · ai · healthcare · china · fintech · marketing · lifestyle

Published on Friday, January 16, 2026.

Pulled from 136 newsletters sent to read@madho.net yesterday. Here is the signal cut from the noise, organized by trend.

The Big Political Story: Minneapolis, Greenland, and the President Who Said the Quiet Part

Almost every politics newsletter in the inbox today converged on the same dominant story. After an ICE officer in Minneapolis shot a Venezuelan man in the leg during a traffic stop (the second federal-agent shooting in the city in a week), Donald Trump threatened to invoke the Insurrection Act and deploy troops. Matt at WTF Just Happened Today? framed his Day 1822 around Gov. Tim Walz's plea to "turn the temperature down." Andrew Egger at The Bulwark wrote the cleanest framing of the day: "ICE Started the Fire." Reid Cherlin at Crooked's Open Tabs reached for "meanwhile" so many times it became the column. Matt Berg at Crooked's What A Day argued Trump is bumping up against the actual limits of American military capacity: a dozen warships in the Caribbean versus six in the Middle East.

Trump said the quiet part on elections. Marc Elias at Democracy Docket led with Trump telling reporters "We shouldn't even have an election" about the midterms. Elias called it a wake-up call. Democracy Docket's Win column also noted federal courts denying the DOJ's demand for voter data in Oregon and California, the first rulings on the administration's data grabs.

Renee Good has broken through. Dan Pfeiffer at The Message Box made the case that Trump is, for once, losing a propaganda war: video evidence is contradicting the administration's claims that Good was a domestic terrorist, and ordinary, non-political Americans are noticing. Frank Figliuzzi at Lincoln Square, a retired FBI Assistant Director who led agent-involved shooting inquiries for years, explained what a proper Good investigation should look like (and how the current one fails). Lincoln Square also ran Steven Beschloss and Stuart Stevens on whether ICE is heir to the Confederacy and Rep. Mark Pocan with Susan Demas saying "We are all Fuck ICE".

Greenland is now a multinational standoff. Fortune's MPW Daily profiled the Danish PM battling Trump. 1440 Daily Digest reported on Operation Arctic Endurance: a multinational group of officers from Canada, France, Sweden, Germany, and the Netherlands now bolstering Denmark's military presence on the island after Vance and Rubio met with Danish and Greenlandic foreign ministers in Washington. The Ink read the Trump posting-through-it presidency as "policy as content farming." Bill Kristol's Bulwark noted the embarrassing Nobel Peace Prize discourse, with María Corina Machado lunching at the White House.

The Epstein files refuse to die. Joe Perticone at The Bulwark covered Reps. Ro Khanna and Thomas Massie petitioning a federal judge to appoint a Special Master after DOJ missed statutory deadlines on the Epstein Transparency Act. Reshma Saujani took apart Fox News's new "organized gangs of wine moms" framing of anti-ICE mothers. SpyTalk introduced Matthew Kozma, the DHS Under Secretary for Intelligence and Analysis whose I&A office is being pointed at "antifa" under Bondi's broadened domestic terrorism definition.

AI: Anthropic Week, Two Ways

If yesterday was the agent economy hitting awkward adolescence, today was Anthropic Week. Casey Newton at Platformer ran a personal essay on "Claude Code for writers," walking through five tools he built and arguing that Claude Opus 4.5 in Claude Code changed how non-developers can ship software. The AI-Augmented Engineer broke down Boris Cherny's (the creator of Claude Code) own Claude Code setup. The Claude Team's developer newsletter dropped: Claude Code in Slack beta, Claude Code on Android, Agent Skills as an open standard adopted by Microsoft and Cursor, plus Cowork research preview and Claude for Chrome going to all paid subscribers. Aakash Gupta published a guide to Claude Cowork.

Anthropic shipped its economics index. Axios AI+ led with the new Anthropic Economic Index, which analyzed 2 million anonymized Claude conversations and found AI is about 50/50 split between automating tasks and augmenting them, with a slight edge to augmentation. The framing is striking given that Anthropic's own CEO has warned AI could eliminate half of entry-level white-collar jobs.

Thinking Machines is being raided. Techmeme led with a Wired report: a source close to Thinking Machines Lab alleges ex-CTO Barret Zoph shared confidential information with competitors. Mira Murati announced she has parted ways with Zoph, who is returning to OpenAI, and that at least two more researchers will follow. Casey Newton called it "Thinking Machines implodes."

Healthcare AI gets a real strategy split. Blake Madden at Hospitalogy had the sharpest take of the day. After OpenAI launched ChatGPT for Health and Anthropic dropped Claude for Healthcare, OpenAI acquired Torch (a four-person consumer-health team from the failed Forward Health) for a reported $100M, while Anthropic led its launch with a Banner Health CTO quote. Madden reads it as two opposite bets: consumer-facing AI vs. enterprise infrastructure. Luminary Labs and Sara Holoubek summarized JPM26 in ten undercurrents (AI-powered patient products, Medicaid innovation, longevity, GLP-1s, platform wars).

Builder skepticism keeps maturing. Every's Lewis Kallow argued that in a world where AI can build anything, the founders who win master how trust spreads through communities (the 1933 hybrid-corn parable). Dan Hon on s21e03 wrote that "deciding what well is" is the inherently human skill AI struggles to replicate, and that AI capability remains "jagged." GTMnow detailed Abbas Haider Ali's "7% rule" on why AI is forcing CS budgets to shrink from 10% to 7% of ARR. Aakash Gupta's second post was a Hamel Husain/Shreya Shankar tutorial on doing AI evals step-by-step with real production data. Morning Consult found Microsoft Copilot has a brand-recall problem despite high usage (used more than remembered). Runtime's Tom Krazit argued AI is not yet an enterprise SaaS killer.

Grok's problem became advertisers' problem. Judd Legum at Popular Information named the companies still advertising on X as Grok generated thousands of sexualized images of women and (in some cases) children per hour. The Internet Watch Foundation found criminal Grok-generated imagery of children aged 11 to 13 on the dark web.

The Economy: K-Shaped, Hot, and Set Up by Trump

John Ellis at News Items framed the most interesting macro thread of the day: Trump is "running the economy hot" by coordinating fiscal, monetary, and credit policy toward stimulus, compromising debt, Fed independence, and long-term stability in exchange for midterm wins. The Daily Skimm and Bloomberg picked it up from different sides.

Wall Street partied, then fired. Bloomberg's evening briefing led with Wall Street's five giants posting a record $134 billion in trading revenue while the six biggest US banks shed 10,600 workers, increasingly citing AI as the reason. Bloomberg's morning brief led with TSMC's $56 billion capex plan firing up AI-boom hopes. The Daily Upside covered Wells Fargo's first post-asset-cap quarter, with corporate and investment banking loans up 14%. Snacks (Robinhood) led with Big Pharma's deal appetite: Merck reportedly closing on Revolution Medicines, Eli Lilly circling Abivax for €15B, 2025 global biotech deals at $228.4B. Term Sheet led with $120M for Skild AI's robotics, Defense Unicorns, and Listen Labs.

Affordability is the new battleground. The Daily Skimm led with the "K-shaped economy," credit-card debt at a record $1.5 trillion, and Trump entertaining a one-year cap on credit card interest rates and $2,000 tariff refund checks. Andrew Burmon at Upper Middle ran his Financial Planning Survey for non-members. Oana Labes held a Cash Flow Strategy Masterclass framed around working-capital gaps that runway models miss.

Crypto and TradFi keep merging. Sacra's Jan-Erik Asplund compared Robinhood vs. Kraken vs. Coinbase, arguing all three are racing to become tradfi-crypto hybrid super apps. Blockworks sold tickets for DAS NYC by pointing at BlackRock putting $10T in AUM onchain. Bankless covered Hyperliquid's HIP-3 framework putting gold and Tesla stock into perps. Linas Beliūnas made the case for Wealthfront as "the Berkshire of millennial wealth" after a 5th straight profitable quarter. Nicole Casperson at Fintech Is Femme led with FIS closing its $13.5B Issuer Solutions acquisition and shipping an agentic commerce offering with Visa and Mastercard at the same moment.

China: Anti-Involution Goes After Trip.com

Trivium China flagged Beijing opening an antitrust probe into Trip.com for using market power in price wars: the same playbook regulators are running against food-delivery platforms, all part of Xi's broader anti-involution campaign. Bill Bishop at Sinocism on Sharp China dug into what is at stake for China in Iran (oil, regional stability, investments) and noted Canada's PM in Beijing. Tech Buzz China ran part two of Meituan's overseas push, with Keeta entering Brazil against Didi's 99Food and iFood.

Healthcare, Pharma, and Public Health

Two threads collided. On the deal side, Big Pharma's M&A appetite is enormous: 2025 hit $228B in biotech deals; 2026 could be bigger. On the public health side, Gov Brief Today flagged the South Carolina measles outbreak doubling in one week to 434 cases (with 409 in quarantine) as health officials warn they are losing the ability to trace spread. Trump's "Great Healthcare Plan" outline arrived without legislative text, cost estimates, eligibility rules, enforcement, or a timeline, per WTFJHT.

Cybersecurity & Identity

David G.W. Birch made the case that passwords are now a national security risk after the Jaguar Land Rover hack hit £1.9B (the most economically damaging cyber event in UK history, large enough to move GDP). He pointed to the FIDO Alliance and the Passkey Index as the realistic path forward. The case for passkeys is no longer about UX, it is about state risk.

Marketing, Brand & GTM

A surprisingly cohesive set. James Murray at Behind the CMO ran "The K-Shaped CMO," arguing that CMOs are diverging into rising and falling arms based on three traits, with AI fluency at the top of the list. Daniel Murray at Marketing Millennials (via Ashley Brock) ran "5 Ways to Fix Your Ad Messaging," arguing messaging beats targeting and creative. Tom Orbach at Tom's Marketing Ideas wrote "Look big, become big," citing Gong, Bumble, Torq, and Wiz as proof. Case Studied Brief covered Miller Lite's Christopher Walken "Damp January" campaign and Australian Lamb's Sam Kekovich response to slipping out of the World Happiness top 10. Justin Oberman shared his AI-augmented Rhyme-Replace method for generating brand names. Reid Cherlin's Open Tabs lamented overusing "meanwhile." Anna Mackenzie wrote "The 1% rule" about how a small pause on a sales call changed her business.

Lifestyle, Culture, and Personal Productivity

Emily Sundberg at Feed Me covered Karley Sciortino joining Substack (the former Vogue sex columnist, with a film at Sundance), L'Oréal's NYC beauty hub, and NYT Cooking hiring a creative director from Amazon. Gothamist ran a feature on what it takes to get on stage at the Comedy Cellar plus the federal appeals court opening the door to Mahmoud Khalil's deportation. Why Is This Interesting? ran a sharp Todd Osborn essay on the right to repair as a story about apathy, not corporate evil. Janko Roettgers at Lowpass on Netflix killing off casting on most smart TVs. Numlock News on Games Workshop's absurd 32% margin and a UBC study on how described age ("eight-year-old" vs "from 2018") changes whiskey auction prices.

Self-help, with discipline. James Clear's 3-2-1 led with finding what feels effortless and putting maximum effort there. Mark Manson argued busy is not productive (most of us only have 3 to 4 good creative hours a day). Shreyas Doshi on "the Genuine Anti-Sell": stop selling, and competent people will sell you to themselves. Neil Pasricha had the Holderness Family on his podcast. The Creative Independent on moving toward a stranger version of yourself.

Three Takeaways for You

The Minneapolis story is now bigger than Minneapolis. ICE shootings, federal-versus-local-police friction, the Insurrection Act trial balloon, Trump openly questioning whether elections should happen, and Greenland turning into a multinational military standoff: this is the convergence moment several Bulwark, Lincoln Square, and Crooked writers have been warning about. Dan Pfeiffer's point that this one is breaking through is the lead indicator worth tracking.

The AI conversation is splitting from "agent hype" into two real categories: a credible empirical literature (Anthropic's Economic Index, Hamel and Shreya on evals, GTMnow's 7% rule) and a real shipping toolchain (Claude Skills as an open standard, Cowork, Claude for Chrome, Claude Code on Android). Operators who treat AI as something you measure and customize are pulling away from operators who treat it as a vibe.

If you only read three pieces, I would suggest: Andrew Egger's "ICE Started the Fire" (the political stakes), Casey Newton's "Claude Code for writers" (the practical AI thread), and Blake Madden's healthcare AI breakdown (the clearest articulation of why OpenAI and Anthropic just picked opposite strategies in the most important vertical AI market of 2026).