Professional yak shaver and web citzen.
Toots in English and "Português" mostly about software development and technology.
| GitHub | https://github.com/augustohp |
| Links | https://about.me/augustohp |
| Pronouns | He/Him |
| Mood | 😌 |
New blog post: "The fate of 'small' open source" https://nolanlawson.com/2025/11/16/the-fate-of-small-open-source/
Some meandering thoughts on writing open source when LLMs can do a lot of the same thing. I don't have all the answers, but I'm trying to ask the right questions anyway.
I'm committed to fully enjoying my weekend but lord I am excited to read this when I have time
Large language models (LLMs) have demonstrated the promise to revolutionize the field of software engineering. Among other things, LLM agents are rapidly gaining momentum in software development, with practitioners reporting a multifold increase in productivity after adoption. Yet, empirical evidence is lacking around these claims. In this paper, we estimate the causal effect of adopting a widely popular LLM agent assistant, namely Cursor, on development velocity and software quality. The estimation is enabled by a state-of-the-art difference-in-differences design comparing Cursor-adopting GitHub projects with a matched control group of similar GitHub projects that do not use Cursor. We find that the adoption of Cursor leads to a statistically significant, large, but transient increase in project-level development velocity, along with a substantial and persistent increase in static analysis warnings and code complexity. Further panel generalized-method-of-moments estimation reveals that increases in static analysis warnings and code complexity are major factors driving long-term velocity slowdown. Our study identifies quality assurance as a major bottleneck for early Cursor adopters and calls for it to be a first-class citizen in the design of agentic AI coding tools and AI-driven workflows.
“Cursor leads to…a significant and persistent increase in static analysis warnings and code complexity…the increase in static analysis warnings and code complexity acts as a major factor causing long-term velocity slowdown”
Large language models (LLMs) have demonstrated the promise to revolutionize the field of software engineering. Among other things, LLM agents are rapidly gaining momentum in software development, with practitioners reporting a multifold increase in productivity after adoption. Yet, empirical evidence is lacking around these claims. In this paper, we estimate the causal effect of adopting a widely popular LLM agent assistant, namely Cursor, on development velocity and software quality. The estimation is enabled by a state-of-the-art difference-in-differences design comparing Cursor-adopting GitHub projects with a matched control group of similar GitHub projects that do not use Cursor. We find that the adoption of Cursor leads to a statistically significant, large, but transient increase in project-level development velocity, along with a substantial and persistent increase in static analysis warnings and code complexity. Further panel generalized-method-of-moments estimation reveals that increases in static analysis warnings and code complexity are major factors driving long-term velocity slowdown. Our study identifies quality assurance as a major bottleneck for early Cursor adopters and calls for it to be a first-class citizen in the design of agentic AI coding tools and AI-driven workflows.
Why do I favour practices like developer testing, TDD, refactoring and continuous integration?
It's to enable rapid & sustained evolution of *working* software.
The odds of delivering the right thing in a single pass are so remote that many iterations are likely required.
And even when we hit the bullseye, it's almost always a moving target. Not least because putting software into an environment has a tendency to change that environment and the problems we need to solve.
The long wait is over!
The #Tusky 24 beta is here!
Check out the release notes: https://github.com/tuskyapp/Tusky/releases/tag/v24.0-beta.1
And how to opt into the beta (or nightly!): https://github.com/tuskyapp/faq/blob/main/README.md#where-can-i-get-tusky
🥳 PHP 8.3 Released!
In this new release, we have:
⌨️ Typed Class Constants
🏛️ Dynamic Class Content Fetch
♻️ A New #[\Override] Attribute
🐑 Deep-Cloning of Readonly Properties
A New json_validate Function
👓 Read all about it on: https://www.php.net/releases/8.3/en.php
🔗 https://php.net/ChangeLog-8#8.3.0
📦 https://php.net/downloads
PHP 8.3 is a major update of the PHP language. It contains many new features, such as explicit typing of class constants, deep-cloning of readonly properties and additions to the randomness functionality. As always it also includes performance improvements, bug fixes, and general cleanup.
Miguel Angel Useche
Nolan Lawson
Cat Hicks
Ben Ramsey
Erik Uden 🚩
J @ Dyne.org
Jason Gorman
Max Goodhart
Tusky
PHP