Juan G.

Exactly

@Bill06988524652 @MagellanQuest Already done, renewable energies are being added as tension control. And BTW, fosil energy producers are sanctioned as they disconnected when blackout events started to happen.

@MagellanQuest They might should look into fixing the, multiple, issues that blacked out the entire country before getting too carried away with what’s in the future.

Spain isn't projecting "a little more" electricity. It's projecting to increase gross production from 273 TWh in 2019 to 422 TWh in 2030. Solar PV: 9% → 138 TWh. Wind: 56% → 130 TWh. Renewables: 37% → 81% of the mix. The bottleneck is no longer energy production. It's the grid, storage, and industrial demand.

I think that @DarioAmodei does not understand software engineering and that he is working feverishly to pump up the valuation of his company in anticipation of its forthcoming IPO.

New research just exposed the biggest lie in AI coding benchmarks. LLMs score 84-89% on standard coding tests. On real production code? 25-34%. That's not a gap. That's a different reality. Here's what happened: Researchers built a benchmark from actual open-source repositories real classes with real dependencies, real type systems, real integration complexity. Then they tested the same models that dominate HumanEval leaderboards. The results were brutal. The models weren't failing because the code was "harder." They were failing because it was *real*. Synthetic benchmarks test whether a model can write a self-contained function with a clean docstring. Production code requires understanding inheritance hierarchies, framework integrations, and project-specific utilities. Different universe. Same leaderboard score. But it gets worse. A separate study ran 600,000 debugging experiments across 9 LLMs. They found a bug in a program. The LLM found it too. Then they renamed a variable. Added a comment. Shuffled function order. Changed nothing about the bug itself. The LLM couldn't find the same bug anymore. 78% of the time, cosmetic changes that don't affect program behavior completely broke the model's ability to debug. Function shuffling alone reduced debugging accuracy by 83%. The models aren't reading code. They're pattern-matching against what code *looks like* in their training data. A third study confirmed this from another angle: when researchers obfuscated real-world code changing symbols, structure, and semantics while keeping functionality identical LLM pass rates dropped by up to 62.5%. The researchers call this the "Specialist in Familiarity" problem. LLMs perform well on code they've memorized. The moment you show them something unfamiliar with the same logic, they collapse. Three papers. Three different methodologies. Same conclusion: The benchmarks we use to evaluate AI coding tools are measuring memorization, not understanding. If you're shipping code generated by LLMs into production without review, these numbers should concern you. If you're building developer tools, the question isn't "what's your HumanEval score." It's "what happens when the code doesn't look like the training data."

@Grady_Booch @ChrSzegedy Exactly, we just changed one language/abstraction/"prompt" for another

This wave is nothing like "higher order programming languges". It's the start of a complete paradigm shift away from using programming languages directly.

The Great Software Quality Collapse techtrenches.dev/p/the-great-so…

@gdb Vibe coding is yet just another incremental rise in abstraction in the history of software engineering

The rise of AI programming agents is changing the nature of software development in the same way as did the introduction of compilers in the time of Grave Hopper. I’ll say it again: the entire history of software engineering is one of rising levels of abstraction.

i changed all our "loading..." states to "thinking.." we are an agentic AI startup now

A friend asked me to look at their code, and *Holy Mother of God!* He works for a medium-sized company. They are building some sort of internal CRM. They've been vibe-coding some of the new features. They are professional developers. Most, I assume, know what they are doing, but I guess it's exhilarating to build 1 week's worth of work in 1 hour and call it a day. The difference between old and new code is clear as day: The old code looks intentional and clean, as if it were written by someone who cared. The new code looks verbose, soulless, and sloppy. Every vibe-coder is now generating as much technical debt as 10 regular developers in half the time. We'll need three generations of developers to clean the mess. We are all becoming rich, for sure.

Personal Update: My p(doom) has gone up. I don’t foresee machines with malice any time soon. But I am starting to see the harm that powerful yet reckless humans who are indifferent to humanity could cause.

El luto que nadie entiende: cuando muere tu mascota jotdown.es/2025/07/el-lut…

One-Factor Authentication

“No puedes convencer a un creyente de nada porque sus creencias no están basadas en evidencia; están basadas en una enraizada necesidad de creer” - Carl Sagan Buenas noches a tod@s

This is probably one of the most stupid things said around AI

I don't understand how someone can be an AI influencer, reposting the same unverified information provided by lying CEOs and VCs and paraphrased by ChatGPT, knowingly fooling thousands of people, not creating any value, simply wasting Earth's limited resources, and still be able to look at themselves in the mirror and have a good appetite.

WTF

¿Qué seguridad proporciona en 2025 firmar CON EL DEDO? Llevo dos años abriendo cuentas bancarias, recogiendo certificados y firmando contratos con dibujitos de Doraemon… y no pasa nada. Eso sí; al notario no le hizo gracia. 😂

results replicated 😬 any comment, @grok? if this has indeed been made part of the sys instructions, we’re gonna need to know why. manipulating search results seems like a funny way to combat misinformation, no? community could use some clarity on this