The most uncomfortable claim here is that expert-built agent skills may become less a path to automation and more a new form of organisational booby trap. Give people a mandate to encode the judgement that makes them valuable, then pretend this is empowerment, and don’t be shocked when the resulting artefact has a few missing teeth. Lee gets the performative bit right because he keeps dragging the discussion back to measurement. A copilot that “saves 40%” but changes neither delivery dates nor staffing nor quality is not evidence; it’s a slide with better lighting. He also lands the point that AI workflows can look cleaner than the systems they replace while being harder to inspect, debug and own. The canvas smiles. The eighth prompt lies on Tuesdays. Yet, I think he overplays the historical analogy, of course. Corporate AI theatre is not famine. But as an organisational warning, it works because the shared mechanism is quota logic: leadership asks for visible production, teams manufacture visible production, and nobody wants to be the first person to say the steel is pig iron. CTOs should be asking dull questions: who owns this after the demo, what baseline did it beat, what happens when the model changes, and which human context did we just delete? Good AI programmes will survive those questions. Bad ones will call them negativity. That’s the tell.

This gets one big thing right: interoperability is usually discussed as a software problem, but the same power game exists in plastic, screws, batteries, chargers, printer cartridges and every awkward proprietary connector hiding in a drawer. F.A.T. Lab and Sy-Lab make that point with toy bricks, which is clever because nobody wants to be the executive explaining why children must respect platform boundaries. The most distinctive claim is also the cheekiest: the Free Universal Construction Kit is “the VLC of children’s playsets”. That’s a small joke carrying a large accusation. This works because the they don’t pretend this is only about play. Lego meeting Tinkertoys is the demo; lock-in is the subject. Yet, a child’s toy is not just an interface. It’s a physical object that can break, splinter, choke, degrade or fail under load. The disclaimers know this, even if the manifesto would rather talk about liberation. Fair enough. Manifestos rarely come with CE testing.

Old games don’t survive because people remember them fondly; they survive because someone packages the emulator, disk image and instructions into something ordinary humans can use. That’s the useful work. Dark Castle nostalgia is the wrapper, but preservation plumbing is the actual story. The site gets that right because it doesn’t merely write a misty-eyed tribute to Duncan, bats and the Black Knight. It gives people a way in. Download, extract, drag DCImage onto Mini vMac, press CTRL-F, and avoid flinging your mouse off the window at a crucial moment. That is preservation with sleeves rolled up. It also captures why Dark Castle mattered: a black-and-white 1986 Mac game that made sound, animation and timing feel distinctive on a little 9-inch machine. Platforms move on, APIs vanish, hardware dies, and culturally important software becomes landfill unless enthusiasts reverse the entropy. Vendors rarely preserve the long tail unless there’s money or brand value in it. Fans do the unglamorous work, then hope nobody with a legal department notices.

The gap between the headline and the evidence is where this one gets interesting. “Inventions increase seven-fold” sounds like the cavalry arriving over the hill. Then the details say something more useful and less comforting: patenting is moving fast, Asia is already ahead, China is accelerating, and Europe has promising niches rather than a commanding lead. That’s a better story. Less triumphant. The EPO gets the central point right because battery circularity is not a nice sustainability garnish. It’s an industrial capability. If batteries become strategic infrastructure, then collection, sorting, recovery, repurposing and refining become strategic infrastructure too. Yet, a patent family is a signal, not a factory. It doesn’t collect a battery, dissolve a cathode, negotiate grid rules, standardise pack design or make hydrometallurgy cheaper on a wet Tuesday. Circularity is being treated as resource policy, trade policy and industrial strategy wearing a green jacket. Europe’s problem is not that it lacks research. The article says it has plenty. The problem is turning patent maps, startups and regulation into scaled plants before Asian incumbents convert their lead into default infrastructure.

A personal observation first: this is the kind of site that makes you realise how much most editorial projects over-explain themselves. Favre doesn’t arrive with a mission statement wearing a sandwich board. Her work leads! For those that follow this newsletter, Malika has been the source of inspiration for the weekly images of the newsletter, but also now the website. What I really enjoyed is her repeatable grammar: bold forms, controlled space, confident colour, clear themes and a strong sense that every piece belongs to the same mind. Obviously that Read Before Monday need its own version of that. Artwork that feels recognisable at thumbnail size. A website that treats every article as part of a larger editorial system. Categories that guide without becoming filing admin. She is a true inspiration! Hope everyone can see that this design cue is not just minimalism. It’s editorial architecture; sharp, spare, recognisable, and just opinionated enough to make the week feel less stupid.