March 26, 2026
https://mariozechner.at/posts/2026-03-25-thoughts-on-slowing-the-fuck-down/
Author discusses about the agents speed of code generation and the increasing amount of sloppy code. As well as how it affects human learning.
Agents are great! They know a lot of stuff. Still, when it comes to coding - we should be capable of evaluating what it does and why it does.
It is fine one off projects, or throw away weekend projects. But when it comes to production grade systems where actual users are visiting and paying for the software - the quality should be at the top level. We should be meticulous in our reviews.
We can use the agents - build repeated stuff or easy stuff so that we can focus on more challenging problems. When it comes to solving a novel problem or architecture change - we first need to think ourselves first before hitting agents up and ask for the solution.
We have to have our home work done. Then we can use agents to refine it further. and finally implement as well.
I also have a concern on Agent written test cases, we should review that more carefully. It should be of good quality.
Instead of blindly auto accepting what the agent spits out, we need to co-work with the agent - check what it does - analyze why it did that - this way we will also build a solid mental model of the code base that will be helpful in future tasks.
Let's see how it goes.
https://nesbitt.io/2026/03/25/the-top-10-biggest-conspiracies-in-open-source.html
Another interesting blog from nesbitt. Well written. Feels like a movie. It will be awesome if this is taken as a movie or an investigative documentary.