Coté logo Coté

For a long time I foolishly scorned George Orwell. I think the prejudice is common among people of my generation (well among people of my generation who care about such things). Orwell is viewed as a fetish of boomer columnists – the types of people who sit in their houses in Hampstead being paid six figures by broadsheet newspapers not really trying with their prose and fondly imagining they are working in plain Orwellian sentences and fearlessly speaking truth to power."

🔗 The mysteries of global shipping, Martin Luther rewired your brain, the AI memorisation crisis, A Walk After Dark and the genius of George Orwell

The US needs, even wants China. From the revealed preferences files:

Aside from rhetorical commitments to restoring domestic industrial capacity and a variety of China Select Committee investigations that tend to go ignored, there’s near-weekly evidence that Congress and the current administration are not entirely serious about leading a whole of nation effort to eliminate regulatory bottlenecks to power abundance, bring pharmaceutical production onshore, mine and refine rare earths, and reduce the leverage that China has over the U.S. economy in about a dozen different areas. Absent solutions on any of those fronts, some of which will take years, Trump couldn’t even ban TikTok. Then, just before Christmas, the U.S. greenlit the sale of high-end Nvidia chips that could help Chinese companies narrow the gap with U.S. competitors in AI. If successfully navigating a prolonged competition with China requires not only acknowledging said competition but actually making the sacrifices necessary to win it, one could be forgiven for reading the news over the past 10 years and concluding that America simply doesn’t have the mettle to prioritize long term national interests over short term private interests (or political incentives)."

Also, pondering a sort of “chilly war” between the US and China.

From Notes from Schrödinger’s Cold War, Andrew Sharp.

SaaS businesses not looking good now:

Traditional estimates for this scope of work: Method Estimate COCOMO (lines-based) ~20,000 hours Feature decomposition ~9,000-12,000 hours Industry benchmark 5-6 years solo, 12-15 months with 5 junior engineers Traditional cost $750,000 – $1,500,000

What I spent: $1,800. That’s a 99.8% cost reduction. It’s also not a fair comparison—traditional development wouldn’t produce identical output. But the delta is large enough that precision doesn’t matter.

🔗 On Running a Startup of Claude Code Agents: What You Get For a Billion Tokens a Month

Outrage about TikTok, so heated a few years ago, now seems like a time capsule from another era. That the negligent, self-dealing new arrangement seems to be going ahead with so little opposition shows not that TikTok has become safer, but that so much else in America has become more dangerous.

From The Economist, gift-link: TikTok is still a danger. America no longer cares

Yes: “about what benefits AI is actually providing today, 66 percent said it’s improving productivity and efficiency.” But: “How that works when only 20 percent report revenue growth is left unanswered.” // And: “Currently, 25 percent of organizations say they’ve shifted 40 percent or more of their AI experiments into live use. That number is expected to reach 54 percent of organizations within the next three to six months.” // Meanwhile, here’s an accounting of executive hopes and dreams.

🔗 Deloitte sees enterprises adopting AI without revenue lift