Links on enumerator.dev

Ten Things I Read this Week

Sat, 01 Mar 2025 00:00:00 +0000

This week’s post is a bit less organized because I was travelling this week.

AI

AI Codegen Workflow

This is a great post outlining a workflow with prompts for using AI to write instructions for itself. Note that Harper switches models for different tasks.

My LLM codegen workflow atm

Use AI to Build AI

Honestly, using AI to build things that build things is the best use of AI I’ve seen so far.

Using Generative AI to Build Generative AI

AI Protest Album

Artists released an album in protest of UK copyright laws changing.

1,000 artists release silent AI protest album Is This What We Want?

AI Breaks the Rules

But what are the rules? The AI broke the rules here because it was determined to win at all costs. I’ve had similar experiences with AI writing code. It is so determined to write code that you must tell it when to stop.

More Research Showing AI Breaking the Rules - Schneier on Security

Engineering Culture

Remember 2004?

This is a great insight into the power shift happening in tech right now.

It feels like 2004 again. - Anil Dash

Normal Engineers

I didn’t read this whole thing because it’s not within my budget to subscribe. BUT! I found this after I wrote Excellent Engineers. Charity Majors always has a fantastic take on work-life balance and engineering culture.

In Praise of “Normal” Engineers 🛠️

Soft Skills

This isn’t the first headline I’ve read like this about AI. Critical thinking and collaboration are skills very different from churning out algorithms. These skills aren’t going anywhere.

OpenAI researcher says soft skills aren’t going anywhere

AI or Die

I hate the title, but it’s a good read about what founders think about AI. AI is changing things, and it’s better to get ahead of it now.

AI or Die | RKG

Random

OMCB

I loved this read about one million checkboxes.

The Secret Inside One Million Checkboxes · eieio.games

Linguists Trick Gemini

One of my new pastimes is reading blogs by linguists who get Gemini to do things it is clearly not good at.

AI Overview jokes

Ten Things I Read This Week

Fri, 21 Feb 2025 00:00:00 +0000

Software Development

Is Engineering Strategy Useful?

There’s always a strategy, even if it isn’t written down.

The single biggest act you can take to further strategy in your organization is to write down strategy so it can be debated, agreed upon, and explicitly evolved.

Is engineering strategy useful?

Hiring

I appreciated this post on an effective hiring pipeline. Put code first and know what you are looking for. Too many hiring pipelines are bogged down with convoluted interviews and challenges that don’t relate to the job.

Hire better, 10X faster: how we screen 50 engineers weekly with no HR—Martian Chronicles, Evil Martians’ team blog

Ethan Marcotte Leaves 18F

Ethan Marcotte, the inventor of responsive design, wrote a thoughtful and harrowing piece on leaving his role at a US government agency.

Moving on from 18F. — ethanmarcotte.com

Leadership

I’ve been reading and writing a lot about the skills needed for our jobs and thinking about how that relates to managment roles. This post gives a good history on the role of middle management in the past few decades.

Geftic predicts there will be new kinds of leadership roles for engineers to pursue that marry strategic planning and technical know-how with the soft skills typically associated with people-management position

Tech’s war on middle management is just getting started

Science & Technology

Catalytic Computing

An attempt to answer, “Can you still use a full hard drive?”

Catalytic Computing Taps the Full Power of a Full Hard Drive | Quanta Magazine

AI

On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?

An academic paper on the risks to society of language models getting larger and larger.

The critical takeaway is that making a language model bigger does not eliminate bias. The size does not equal diversity.

If we want sharper LM tools, we need sharper training data.

https://dl.acm.org/doi/10.1145/3442188.3445922

AI Is Not About “Prompt Engineering”

Shocker. I know. This is a great article about using LLMs for software development. In my own words, learning to use LLMs means learning the skills we’ve always needed for the job.

Using LLMs effectively isn’t about prompting

Brigitta Bökeler on Reasoning and Coding

Reasoning models are still too slow to be helpful in coding tasks. I’ve found this to be true myself. At times, it takes just as long to give a model the right context and wait for its response as it would for me to try a solution out myself.

Exploring Generative AI

Random

Life in Weeks

Gina Trapani mapped out her life in weeks. But more importantly, I looked at the HTML and this is also her life in weeks as HTML buttons.

My Life in Weeks by Gina Trapani

On Fixing

I was reminded of Norman White’s post about the right to repair when I was researching for Good Hacks

I don’t really fix toasters, although I’d be proud if I could. Almost nobody fixes toasters. This is because a modern toaster is nearly impossible to fix, held together with little bendy tabs which break off if you bend them more than twice. The toaster manufacturer naturally expects that you do the Right Thing – toss that dysfunctional item in the dump and buy a new one! All in all, the working toaster is a perfect symbol for modern utility in general… glamourous and efficient! Nevertheless, staring at this glamourous efficient high-resolution computer screen for hours at a time, you and I are both wrecking our eyes, not to mention our social lives. But, hey, I don’t mind… do you?

https://normill.ca/onfixing.html

Things I Read This Week

Sat, 15 Feb 2025 00:00:00 +0000

On Engineering

Chip Huyen on Pitfalls When Building Gen AI

This is a great post if you’re asking “should I use AI for that?”

Use generative AI when you don’t need generative AI

This pitfall is so common when technologies are new. I love Chip’s example because using AI to save on electrical bills is like using a diesel generator to charge your EV.

Common pitfalls when building generative AI applications

Vibe Coding

I hate it.

This has also made me think that vibe coding existed before LLMs were widely accessible. We just called it “legacy” code.

There’s a new kind of coding I call “vibe coding”, where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It’s possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper…

— Andrej Karpathy (@karpathy) February 2, 2025

On Security and Critical Thinking

Pairwise Authentication of Humans

Ever worry an AI replicant is calling you? Yeah, me too. You and your friends can generate OTP codes to verify if you’re real.

Pairwise Authentication of Humans - Schneier on Security

When the axe came into the forest, the trees said: the handle is one of us.

When the LLM came into engineering the developers said: hey, it’s made of code!

Quote Origin: The Trees Voted for the Axe Because the Axe Handle Was Made of Wood – Quote Investigator®

On Technology & Society

First distributed gate teleportation

I know next to nothing about quantum computing. But this was interesting. distributed gate teleportation means we could soon do large-scale quantum computing.

Oxford team teleports first quantum gate in landmark paper

We’re Already at Risk of Ceding Our Humanity to AI

The idea that machines could or should take over making art or writing diminishes the work of teachers, of human-authored books about how to make art and to write, and the fact that people can learn stuff.

We’re Already at Risk of Ceding Our Humanity to AI