"AI-generated code will choke delivery pipelines" just published. Algorithm-friendly link in reply.

A good illustration of part of a staff+'s scope of work is stopping initiatives like this one. Also of its challenges, as it doesn't make friends.

I chose the green door ninety-three days ago. At the time, it seemed obviously correct. Not even a close call. The red door offered two billion dollars immediately—a sum so large it would solve every material problem I'd ever face, fund any project I could imagine, and still

Some of the best stuff on the internet is on Reddit

https://t.co/HqxebZors7

AI applied to sw eng has this basic theme: The speed increase amplifies output, but also every known challenge (distributed systems, software delivery bottlenecks, tech debt, security ... ). And most of the latter have compounding effects.

We've been hiring engineers for the last 7 weeks and the consensus on my team is that this is the hardest engineering recruiting cycle ever. It's like AI has rotted people's brains. We routinely interview experienced-on-paper candidates who know nothing and are unable to code.

The externalization of Google's engineering culture has led to incredible amounts of complexity spread across the industry.

Reading the AWS postmortem, and the HN reaction to it, and: I will never stop evangelizing this, the most brain-altering piece of writing I've encountered.

While Generative AI can increase the overall volume of documentation and code, in terms of curated and validated release content, the increase might only be +20%. The following represents the number of lines of ~6 months of work on similar projects, both before and after using AI

Daniel is closing in on measurable impact over the industry’s productivity

YAGNI has its nuances too

"Utility and value are distinct concepts."

@bongocourt Yep! A lot of Stripe is trying to make the financial system work the way people think the financial system works.

At least it is one of the few covers that rival the original

https://t.co/8ztagpCMZp

Abundance works and Austin is ground zero. You can literally just build more housing.

Saying that deep learning is "just a bunch of matrix multiplications" is about as informative as saying that computers are "just a bunch of transistors" or that a library is "just a lot of paper and ink." It's true, but the encoding substrate is the least important part here.

The real point here is not whether ApproachA is capital-B better than ApproachB. It is that what actually matters is having good judgement to decide when to apply A or B based on context. That judgement being a type of knowledge that isn't "technical"

This opposition is common but we always get the angle wrong. What you get aren't "counter examples". This is not a mathematical proof aspiring to absolute validity. You get contexts in which $approach just happens to be worse, but that coexist with contexts where it is better.