Monthly Archives: April 2025

Source: OSNews

I love the Kate Text editor. I use it for pretty much all the programming projects I do. Kate has been around for long time now, about 20 years! At least earliest blog post for it I could find was written in 2004.

I wanted to go over my workflow with it, why I like it so much and hopefully get more people to try it out.

↫ Akseli Lahtinen

Programmers and developers tend to be very set in their ways and have their preferred workflows – which profession doesn’t, honestly – and since there’s such a wide variety of developer and programming tools out there, it feels like every single developer’s workflow and setup is entirely unique. Akseli Lahtinen, KDE developer and allround awesome person, details his setup using Kate, the venerable and feature-rich text editor from the KDE project.

As someone who can’t program, I can’t really compare his workflow to my own, but what I found interesting while reading his post is that there’s quite a bit of overlap between my previous work as a translator and his work as a developer. While the contents of each individual view inside his Kate window are obviously different, the setup of windows and tools I had when translating looked very similar.

This shouldn’t be surprising to me – after all, both translating and developing requires multiple work surfaces, language plugins, formatting tools, tons of keyboard shortcuts, and a whole load of browser tabs, PDF files, and other documents to find just the right translation or the perfect term, as well as a ton of background to make sure you understand the topic you’re translating about. Y’all have no idea how much I know about the deepest complex inner-workings and processes of some of the largest organisations in the world, just because I needed to study them and had access to their internal documentation and software.

I also read and studied way too many complex contracts, European law, and technical studies into medicine and healthcare treatments, and I guess developers and programmers do the same thing – just focusing on different subjects. What’s the best way to do this thing in the programming language I’m using? How does this library I want to integrate work? What are the API endpoints for this service I want to use?

It’s really not that different from translating, and that never really dawned on me until now.

Source: Slashdot

While college applicants are often required to write a personal essay for their applications, political scientist/author/academic Yascha Mounk argues that's "a deeply unfair way to select students for top colleges, one that is much more biased against the poor than standardized tests." The college essay wrongly encourages students to cast themselves as victims, to exaggerate the adversity they've faced, and to turn genuinely upsetting experiences into the focal point of their self-understanding. The college essay, dear reader, should be banned and banished and burned to the ground. There are many tangible, "objective" reasons to oppose making personal statements a key part of the admissions process. Perhaps the most obvious is that they have always been the easiest part of the system to game. While rich parents can hire SAT tutors they can't sit the standardized test in the stead of their offspring; they can, however, easily write the admissions essay for their kid or hire a "college consultant" who "works with" the applicant to "improve" that essay. Even if rich parents don't cheat in those ways, their class position gives rich kids a huge advantage in the exercise... [W]riting a good admissions essay is to a large extent an exercise in demonstrating one's good taste — and the ability to do so has always depended on being fluent in the unspoken norms of an elite community... Many on the left oppose standardized tests on the grounds that they have a class bias, and that hiring a tutor can make you perform better at them. But studies on the subject consistently suggest that the class bias of personal essays is far stronger than the class bias of standardized tests.... But the thing I truly hate about the college essay is not that it is part of a system that keeps deserving kids out of top colleges while rewarding privileged kids who (to add insult to injury) get to flatter themselves that they have been selected for showcasing such superior personality in their 750-word statements composed by their college consultant or ghostwritten by ChatGPT... [W]hat I truly hate about the college essay is the way in which it shapes the lives of high school students and encourages the whole elite stratum of society — including some of its most affluent, privileged and sheltered members — to conceive of themselves in terms of the hardships they have supposedly suffered... [I]t is the bizarre spectacle of those kids from comparatively privileged backgrounds being effectively coerced by the admissions system to self-exoticize as products of great hardship which I find to be truly unseemly... And this is why I suspect that the seemingly innocuous institution of the college essay is more deeply damaging — to the high school experience, to the self-conception of millions of Americans, and even to the country's ability to sustain a trusted elite — than it appears... [I]t drains the souls of teenagers and encourages a deeply pernicious brand of fakery and breeds widespread mistrust in social elites. The college essay is absurd and unfair and — ironically — unforgivably cringe. It's time to put an end to its strange hold over American society, and liberate us all from its tyranny.

Read more of this story at Slashdot.

Source: Ars Technica

When it comes to actually storing the numerical weights that power a large language model's underlying neural network, most modern AI models rely on the precision of 16- or 32-bit floating point numbers. But that level of precision can come at the cost of large memory footprints (in the hundreds of gigabytes for the largest models) and significant processing resources needed for the complex matrix multiplication used when responding to prompts.

Now, researchers at Microsoft's General Artificial Intelligence group have released a new neural network model that works with just three distinct weight values: -1, 0, or 1. Building on top of previous work Microsoft Research published in 2023, the new model's "ternary" architecture reduces overall complexity and "substantial advantages in computational efficiency," the researchers write, allowing it to run effectively on a simple desktop CPU. And despite the massive reduction in weight precision, the researchers claim that the model "can achieve performance comparable to leading open-weight, full-precision models of similar size across a wide range of tasks."

Watching your weights

The idea of simplifying model weights isn't a completely new one in AI research. For years, researchers have been experimenting with quantization techniques that squeeze their neural network weights into smaller memory envelopes. In recent years, the most extreme quantization efforts have focused on so-called "BitNets" that represent each weight in a single bit (representing +1 or -1).

Read full article

Comments

Source: Hacker News

Source: Slashdot

An anonymous reader shares a report: Members of the European Parliament were offered special pouches to protect digital devices from espionage and tampering for a visit to Hungary this week, a sign of rising spying fears within Europe. Five lawmakers from the Parliament's civil liberties committee traveled to Hungary on Monday for a three-day visit to inspect the EU member country's progress on democracy, the rule of law and fundamental rights. One lawmaker on the trip confirmed to POLITICO that the Parliament officials joining the delegation were offered Faraday bags -- special metal-lined pouches that block electromagnetic signals -- by the Parliament's services and were also advised to be cautious about using public Wi-Fi networks or charging facilities.

Read more of this story at Slashdot.

Source: Ars Technica

The verdict is in, and Google has been found to illegally hold online ad tech monopolies.

For over a decade, "Google has willfully engaged in a series of anticompetitive acts to acquire and maintain monopoly power in the publisher ad server and ad exchange markets for open-web display advertising," tying its publisher ad server and ad exchange together "to establish and protect its monopoly power in these two markets," the ruling said.

At trial, the DOJ argued that Google's ad business expanded to choke out competitors and benefit only Google. They argued that Google "rigged" ad auctions, allegedly controlling "multiple parts" of services used to place ads all over the Internet, unfairly advantaging itself in various markets.

Read full article

Comments

Source: Hacker News

Source: Schneier on Security

As AI coding assistants invent nonexistent software libraries to download and use, enterprising attackers create and upload libraries with those names—laced with malware, of course.

EDITED TO ADD (1/22): Research paper. Slashdot thread.

Source: The Register

Buy high, sell low: FPGA biz cost x86 giant $16B decade ago

A decade after gobbling up Altera, Intel is loosening its grip. On Monday, the x86 giant said it's flogging a 51 percent stake in the FPGA slinger to private equity firm Silver Lake.…

Source: Hacker News