Notes

To bypass the issue of grid demand, many data center builders have promised to power the buildings themselves. The problem with that plan is that it’s easy to build a warehouse full of GPUs. Building power plants and transmission lines to power those warehouses will take years and cost billions more than the data centers. A skilled and efficient builder can complete a data center construction in under a year. Building new energy generation to meet that data center's power needs could take a decade.

When it comes to building new power plants and building out the grid, nothing is ever as fast and easy as one would expect it to be.

Most of these turbines are across state lines in Mississippi, are operating without public permits and are destroying the air quality of people who live near them. Some of the gas turbines weren’t disclosed. A lawyer for the Southern Environmental Law Center told Reuters the turbines are a violation of the Clean Air Act. It may not matter. On July 27, Trump’s Environmental Protection Agency announced that power plants that exclusively keep data centers online won’t be subject to the Clean Air Act.

Much like everything else this administration has done, destroy the planet to further enrich the rich and powerful.

At the heart of this and as I mull this over again and again is the idea that the requirement to earn a profit leads to behavior that's directly counter to the open ideals upon which the web was founded. Removing that requirement allows for more openness (though that is still voluntary). If you believe in the open web, you'll make things open. If you support the ideal, you'll build for it. If all you see is a platform and a medium absent the ideal, you'll exploit it.

It's the ongoing debate about capitalism versus community. Too often the lust for more and more profit ruins good things.

In the case of software, the industrialisation of production is giving rise to a new class of software artefact, which we might term disposable software: software created with no durable expectation of ownership, maintenance, or long-term understanding.

I can see where spinning up a quick prototype might be a good use of vibe coding. Maybe even a little personal-use app could as well. But I would hate to see well built, and well designed applications become shells of their former selves.

The open question, then, is not whether industrial software will dominate, but what that dominance does to the surrounding ecosystem. Previous industrial revolutions externalised their costs onto environments that seemed infinite until they weren't. Software ecosystems are no different: dependency chains, maintenance burdens, security surfaces that compound as output scales. Technical debt is the pollution of the digital world, invisible until it chokes the systems that depend on it. In an era of mass automation, we may find that the hardest problem is not production, but stewardship. Who maintains the software that no one owns?

The main winners in a gold rush are the ones selling pickaxes, and it sure seems to me like the token vendors are about the only ones who really stand to gain from most of this, if we’re adhering to any coherent, holistic definition of productivity.

I wonder how much of the supposed productivity gains are either eaten up by reviewing the pull requests, or later, fixing the tech debt and bugs caused by the code generators. But if the AI companies can convince your boss they can get more work out of you, whelp, there's the ball game.

In one article on its site, ISBNdb explains that printed books published before 2022 are ideal for AI training data because they don’t include AI generated text.

This just makes me sad. I know it's nothing new, but seeing books destroyed to create new AI slop is just painful to see.

When everyone fakes an opinion to look engaged, you end up with a five-way debate over something that should’ve been a coin flip. Real opinions get drowned out by performative ones. “I don’t care, just pick one” is permission for the room to move on

There's not always a single correct way to build things. Choose one and go with it. You will likely be able to change it later if the need arises.

Some things deserve your conviction. A lot of things don’t. Knowing the difference is the skill.

That's the key part, there are some thoughts and decisions that are more important. Focus your efforts there.

In Texas just this Tuesday, an ICE officer shot and killed Lorenzo Salgado Araujo, a father of three who lived in the US for 35 years building houses and caring for his family. The agency immediately released a statement justifying lethal force on someone it alleges tried to “weaponize his vehicle.” Videos showing parts of the confrontation already suggest it’s probably another bullshit story like the ones we’ve seen from Minnesota. The feds later admitted they were looking for an entirely different person. Salgado Araujo nonetheless ended up dead.

ICE will never tell the truth about what they do. They don't want accountability. The media and everyone needs to realize this before they start printing and believing more ICE bullshit.

The government is very upset that people are criticizing it for shooting innocent mothers to death in the street. DHS stalked and intimidated a man who protested against state-sponsored killing, issuing him a “WARNING NOTICE” as flimsy as the nonjudicial “warrants” it uses now to bust into people’s homes. It’s like if Jay and Silent Bob showed up at your house because of an online comment you wrote, except with guns and the force of the federal government. It’s both deadly serious and deeply unserious. What are they afraid of? Well, we already know. Accountability in any form.

I guess the First Amendment doesn't matter if you hurt their feelings.

Policing in America has always been complicated, but the Trump administration has made it a runaway train of abuse fueled by billions of dollars and unapologetic racial animus. Immigration and Customs Enforcement has gotten so much money and permission from Trump that it operates its dear leader’s fundamentally racist mission of mass deportation with impunity. ICE agents have kidnapped people, knocked down doors without warrants, assaulted reporters, terrorized the general public (including children and infants), and even shot dead innocent people in broad daylight. And that’s just the brief summary of events.

ICE is a menace to democracy and its murderous conduct should be soberly considered. But we should also recognize that these people are weak and sad and ought to be made fun of. Even for their stupid outfits, which look ripped from a “how to be tacticool” buying guide. And Americans are ridiculing these people. What better way to protest an army of clowns than by showing up as Portland did with a human frog at the front? Cops arrested that frog. Trillions of dollars later, we’ve learned nothing about counter-insurgency, because now there’s a frog legion.

These ICE agents are so small that they wear masks in public like a bunch of cowardly Patriot Front wannabes, who probably got a similar amount of training. You’d think the average DMV worker would have been masked up for years based on the abuse they get, but then again, the average government employee is far braver than these fools.

I don’t miss floppy disks. I miss the mindset they demanded—that every byte matters, that constraints breed creativity, and that software should be light on its footprint.

I remember booting up the Amiga computer as a kid using floppy disks. All the games I played on it used 3.5-inch floppy disks and I don't remember any of them needing more than 1 disk.

Anthropic ran a randomized trial in early 2026 where engineers learned a new Python library, half with AI assistance and half without. Both groups finished the tasks at the same speed. But the AI group bombed the follow-up comprehension quiz: 50% versus 67% for the manual group, with the gap widening on debugging. The interesting cut was inside the AI group itself. Engineers who used AI to ask conceptual questions scored above 65%. Engineers who copy-pasted the generated code scored under 40%.

During a televised cabinet meeting at the White House in 2025, Musk, wearing a murdered-out MAGA hat signed by his boss, had a giggle about “accidentally” canceling Ebola prevention. He said it was a mistake that would be fixed. USAID whistleblower Nicholas Enrich, testifying before Congress, said that fix never came. A little more than a year after Musk’s comments, Africa is facing what could become the worst Ebola outbreak ever.

Cause and effect seems to be a concept that is so often lost.

You might think Musk’s actions gutting effective global health programs clash with his downright creepy quest to raise birth rates, but not when you consider the guy is also a huge racist. The list is too long to keep score, but a few relevant highlights: He’s stoked claims of a “white genocide” in South Africa, spent most of January posting other white supremacist talking points, and most recently has been encouraging race riots in the UK all to gin up anti-immigrant sentiments. Is there any doubt about why this guy had so much fun destroying one of the most successful global health initiatives in history, which saved millions of Black people?

Musk and the current administration aren't even trying to hide their racism anymore.

Musk hijacked the government to destroy these missions and dehumanize these people in service of the total lie that it would make the government more efficient. Of course, there’s a big difference between “efficiency” and incompetence. An agency isn’t more efficient if it doesn’t exist; it’s simply been murdered. A fire department with no firefighters looks good on a balance sheet if you can ignore that the city is ablaze.

And here's the bit that really keeps me up at night: a lot of this AI-generated code? Nobody fully understands it. The person who "wrote" it didn't really write it. They prompted it, skimmed it, maybe ran it once. When it breaks in production at 2am, the person on-call didn't write it and the person who prompted it can't explain it. You've just increased the surface area for incidents while decreasing the number of humans who can reason about the system.

“Another developer on a contract working with me at the moment generates massive amounts of code, leaving me with 1000+ lines of pull requests to review and it takes massive amounts of time to do this. This leads to me feeling more tired and burned out than I've ever felt in my entire life. The cognitive overhead of switching between prompting, coding, checking the LLM's output is a massive energy drain. It has not been a productivity booster at all, it feels like a speedrun towards severe mental exhaustion.”

There's a reason why large pull requests are frowned upon. They're harder to review even when the developer knows all the code that is there since they wrote it. Now devs are being asked to review massive pull requests with code that LLMs wrote that may or may not have even been looked at by the developer submitting the pull request. It feels like a recipe for disaster.

The problem all the developers I talked to agreed on is that the more they relied on AI to code, the more the skills they’ve honed for years deteriorated. This is by now a well studied phenomenon sometimes referred to as "cognitive debt” or "cognitive atrophy.” The idea is that people who use AI to automate certain parts of their job lose the ability to do those tasks well, therefore de-skilling themselves.

So now us devs have to continue to keep up with all the changes in tech stack and best practices, but we have to keep the AI from eating at our ability to do the jobs we were trained to do, and well.