RSSAmplifier

Blog

Frédéric van der Essen

Essays by Frédéric van der Essen

frederic.vanderessen.comRSS feed ↗8 posts

Latest posts

Unsupervised Codebase Refactoring: How To, What Works and What Doesn't

Could code refactoring become a one-click operation, like running a formatter but for architecture ? Split the bloated files, untangle the responsibilities, deduplicate the logic, and come back to a clean codebase ? So I ran an experiment. I pointed Opus 4.6 at a 60K line Go microservice I had vibe coded over the past few months, gave it some refactoring principles, and let it run unsupervised.…

My Programming Job Has Become An Intelligence Buying Job

I have been programming for more than 20 years. 20 years of learning how to translate thought into code. 20 years of trying to master an editor to be fast at it. 2025 was the year when I wrote my last piece of code. AI is now better at it than I am and ever will be. It is not just better at writing code, it is better at architecting, debugging, designing, documenting, deciding trade-offs. It is…

AI Writes Code Fast, Human Reviewers Can't Keep Up

The diagram above approximately represents the process I have to go through to complete a programming task at work. This is the result of using the standard workflow that GitHub and Gitlab recommend. Which means picking up an issue, making some code changes, creating a merge request for these changes, and getting the CI checks, and approval from my colleagues, before it is merged and automatically…

The Human Only Public License

Whether artificial intelligence systems will end up being a positive or a negative force for humanity is still an open question. But we might find ourselves one day with AI embedded at every layer of our existence, living lives of toned down and diluted humanity with only our dreams for escape. Although I am not yet convinced of this worst case scenario, I believe it is important that we as…

Never Ever Use Content Addressable Storage

Content Addressable Storage (often abbreviated as CAS) is a technique to derive the id of a document from its content, basically by taking its hash. Imagine you want to store pictures in a file store, and you need an id to reference that picture, you simply take its hash. This technique has benefits, first, if you have a document you automatically know its id and can find other references to it.…

What is Good for AIs is Good for Humans

I have discovered something marvelous; whatever you do to make the AIs more effective at helping you do your job is also going to help your fellow human colleagues. Let me take an example; comments in code. I have been guilty of never putting much comments in my code. The reason being that I didn't need them (I wrote the code, I know what it does), and that they're a pain in the ass to maintain.…

GPT-4 Can Use Tools, Serve a REST API and Pilot Drones

Note: I wrote this article before OpenAI's announcement of their plugin integration which uses similar techniques to the ones highlighted in this article (things move fast these days ...) but I believe it is still worth a read ! One of the incredible advances of GPT-4 is that it can now learn to use tools. By tools, I mean that it can learn how to gather information from outside its embedded…

Better Javascript Promises

This is a repost of an old article from 2014 that was lost in a blog migration but I think is still relevant. It shows how differently the javascript promise problem could have been solved. This was written before async await became a thing, and before the function coloring problem was given its name. This proposal would have solved the promise problem without the coloring, but alas, it wasn't…