Technology

Technology is treated here as material to be taken apart, measured, and put back into service, not as a stream of announcements to comment on. This category holds both of the blog’s registers: the honest bench test, a keyboard, a vertical mouse, an OLED panel pressed into service as a monitor, a domesticated electricity meter, a server running FreeBSD rather than Debian, video encoding measured with VMAF rather than judged by eye; and the critique of what hardware and platforms do to their users, from the dongle that refuses to talk to its neighbor to the service you thought you owned that changes the rules mid-course. You will find pieces written after real, long-term use, often revisiting something that once seemed like a sound purchase, and a lasting interest in old ideas that still hold up, from the BGP protocol to the X Window System. Written by a practitioner who prefers a justified technical choice to a defended preference, and who provides the numbers when he has them.

Salvation by Compute: The Promise of Defeating Aging

Philosophy

Techno-optimist discourse no longer promises tools. It promises deliverance: the end of disease, and then of aging itself. This essay does not try to say whether that promise will hold, but examines its shape, which borrows the grammar of religious eschatology. Noting the resemblance is not enough to discredit a claim, on pain of committing the genetic fallacy. What separates science from belief here is not the content of the promise but its relationship to evidence: what result, what failure, would make you give it up? One of the great forms of faith of our era is taking shape in front of us, and it is being written entirely elsewhere.

China builds machines, France sells shares

Digital sovereignty

On July 28, 2026, China began volume production of its own immersion scanners, the 60-million-euro machines that print circuits onto silicon and that only ASML truly knows how to build: five units this year, about twenty in 2027, against ASML’s 130 systems a year. One to twenty-five: the ratio looks trivial, yet Seoul dropped 10% and Tokyo 4% in the days that followed, because a technology embargo has no proportional value. Its value is binary, and it has just stopped being a wall and become a delay, something that can be calculated, amortized, and planned around. By removing the alternative, export controls created the competitor they were meant to smother, exactly as they had taught Chinese AI frugality. Three days later, Bpifrance sold 2.5% of Orange for 1.1 billion euros and called it asset rotation: for the same sum, one bought productive capital, the other three months of cash.

Jensen Huang Makes the Case for Open AI Models… and for Our Sovereignty

Digital sovereignty

For his very first post on X, Jensen Huang said nothing about GPUs: he shared a letter defending open models, signed by NVIDIA, Meta, Mistral and Hugging Face, but conspicuously left unsigned by OpenAI and Anthropic. That list of absentees says a great deal: the line between those who signed and those who held back maps precisely onto the line between an economy built on openness and one built on the tollbooth. The text speaks of “American leadership,” yet every one of its arguments makes the case, in spite of itself, for everyone else’s sovereignty. Because an open model knows no borders: the same file that spreads AI through the factories of Ohio also lets you run inference at home, out of reach of the Cloud Act.

From WordPress to Astro: The False Simplification of the Modern Web

Softwares

Switching from WordPress to a static generator like Astro is frequently framed as the ultimate move toward simplicity and high performance. However, replacing a single, consolidated dashboard with Markdown files, Git workflows, and fragmented cloud services shifts technical complexity rather than removing it. This transition creates organizational bottlenecks for non-technical teams and exposes projects to npm supply-chain vulnerabilities instead of traditional CMS threats. Furthermore, relying entirely on proprietary cloud stacks trades open-source flexibility and digital sovereignty for vendor lock-in. Ultimately, for standard marketing websites, a battle-tested, self-hosted CMS remains the most practical and reversible choice.

The Claude Fable 5 Illusion: You Pay for 20x, You Get Rationed to 6.7x

Technology

On July 20, Anthropic folds Claude Fable 5 into its Max and Team Premium plans at 50% of limits, on the very day those base limits drop by about a third. Compound the two cuts, both publicly announced, and a Max 20x subscriber goes from a nominal 20x to effective Fable access of roughly 6.7x, a two-thirds erosion of the promise. This is not a hunch; it is arithmetic on official figures. While Washington keeps its grip through the Cloud Act and a rival undercuts on price, Anthropic keeps the premium badge on the invoice while rationing the engine under the hood. Welcome to shrinkflation applied to tokens.

Anthropic doesn’t know how Claude works, and that’s excellent news

Artificial Intelligence

Anthropic just published the strangest discovery of the year: an internal structure in Claude, essential to reasoning, that nobody designed and that emerged on its own during training. A company valued at nearly a trillion dollars had to invent an entire science, interpretability, to understand its own product after the fact. That’s not a confession. It’s the cleanest experimental demonstration ever produced of a Hayekian theorem: you can write the rules of a system without being able to know the order that will emerge from them. And while the tech press marvels at finally being able to read the machine’s thoughts, nobody is asking the two questions that matter: what will the AI Act’s explainability requirement be worth against knowledge the creator itself can only obtain through research, and who will hold the lens that can not only read those thoughts, but edit them without leaving a trace?

GD, ImageMagick, libvips: the image engine, that decision we make by default

Softwares

Most PHP developers never chose GD, it was simply there, compiled in by default, and the question of the image engine ended there. That is a mistake, because this invisible component decides the speed of your back office, the memory bill of your server, and the quality of the images you serve, especially once a media library has to absorb batches of heavy photos. Against GD and ImageMagick, libvips changes the game through its streaming architecture: where the others load the whole image into memory, it works in strips, which makes it several times faster and far more frugal on large sources. The numbers bear this out, from the official benchmark to production reports, with one honest caveat: on very small images, its advantage disappears. This article compares the three engines, explains why libvips stayed invisible to the PHP world for so long, and why the official Intervention Image v3 driver is finally establishing it as the deliberate default for the hot path.

Local AI: you don’t pay twice, you pay three times

Artificial Intelligence

The standard line is that local AI costs you twice, the machine and then the watts, and that’s true. But there’s a third bill, the only one not denominated in euros: the bill for sovereignty, which you sign silently the moment you decide to “ride it out on APIs” for two years while waiting for prices to fall. On top of that, the French watt is among the cleanest on the continent, and the memory spike you’re advised to wait out isn’t weather but a perfectly rational eviction, with no reason to correct itself on its own. A companion to tazmenworld’s buying guide, adding the line the spreadsheet forgets and a reminder that, in a world where memory goes first to wherever the margin is fattest, owning your own is already an act of resistance.

Logitech PRO X 2 on Mac : The wireless promise that was never kept

Economy

For two years now, my Logitech PRO X 2 headset, connected over Lightspeed to an Apple Silicon Mac mini, has been dropping out without warning, in total silence, taking the machine’s entire audio subsystem hostage along the way. Cross-referencing other users’ reports, the conclusion is beyond dispute: it is neither my bad luck nor my setup, but the collision of two fragilities, Logitech’s proprietary wireless and the macOS USB audio stack, made worse by its G Hub software layer. The January 2026 expired-certificate fiasco, which paralyzed G Hub on macOS alone, is the glaring proof. The Bluetooth fallback offers relative stability, but at the cost of latency and quality, which is to say a confession of failure dressed up as a feature. At 300 dollars, that is a disappointment that has cured me of the Astro A50 X and made me hesitate over my future MX Master 4.

Has Apple finally fixed the macOS SMB Disaster?

Operating systems

Against all odds, Apple has finally responded to a two-year-old Feedback Assistant ticket regarding the abysmal performance of the SMB protocol on Mac. According to the company, beta 2 of macOS 27 Golden Gate delivers a long-awaited fix for these chronic slowdowns and disconnections. This update is particularly crucial, as this latest release marks the definitive retirement of the AFP client, forcing all NAS users to fully transition to Microsoft’s protocol. It remains to be seen whether Apple’s promises hold up to real-world crash testing or if power users will still need to rely on makeshift configuration files.