Writing

Things I
think about

Notes on how the work is changing — and on what hasn't.

Value is moving from execution to judgment

AI has not made thinking less valuable. It has made thinking the job.

A developer at a desk: on one side long hours, manual work and repetitive tasks; on the other, defining the problem, asking the right questions and making better decisions
Read the full article

One of the most profound shifts I have noticed in the age of artificial intelligence is that value is no longer defined by execution alone. Instead, it is increasingly defined by how we think and the decisions we make.

Why I used the Hungarian algorithm

A greedy matcher saved 47 in cost and left a client with nobody. Every decision it made was the cheapest one available.

Two cost matrices compared: greedy fills 7 of 8 requests at a total cost of 605, the Hungarian algorithm fills 8 of 8 at 652
Read the full article

When I started building benchFlow, a system that decides which engineer goes to which client request, I did the obvious thing first.

Old habits are not best practice

Git never stopped us committing a .env file. We avoid it for a good reason — it is worth checking we still remember what that reason is.

A .env file behind a padlock beside a Git changes list, contrasting the old habit of sharing .env files by hand with committing a .env.example and keeping the real secrets in a secret manager
Read the full article

What a developer said about Git and .env files caught my attention — especially the idea that we sometimes stick to certain development practices not because they are the best, but simply because we are used to them.