I often realize it when I find it strange that something just works and does the thing it should be doing. You had one job and you did it well. How amazing and rare ;) That shows you how low our expectations are.

I'm just going over this with a friend that has a great idea for a medical software app :D I think we should get better at explaining people how expensive and difficult custom software is. It should always be the last resort, after exhausting all other options IMO. Unless... you're a dev yourself :P

It's easy to find problems in many software products, but every time I make a withdrawal in Deel (like today), it feels so good it's almost surreal. Looks so obvious, just "as things should be", yet for years I had so many challenges with various systems

An image worth 10'000 word essay ;)

"Things that cannot happen... happen. All the time" Today is one of those days when that reality hits really hard ;) In my experience, this is one of the major reasons why working with legacy systems is so challenging and interesting at the same time.

That conversation didn't dissappoint, especially on points where there was some disagreement between the panelists :) My key takeaways: Zitron - many "useful AI" examples are not really LLMs, or where there before the current wave of hype, they're the "old school, different kind of AI" rebranded.

Trade-off - a decision between two GOOD options. I read it a while ago, and it keeps coming to my mind very often. I realized that saying yes was often easy, but many important decisions were about what I was ready to say no to. So true in life, so true in software.

As a Kanban/Lean fan I think that the concept of "muda" is something painfully missing from Scrum and other Agile approaches in software. It's always more, faster, better until it becomes overwhelming and exhausting. But sometimes you can gain so much more by reducing waste and tiny improvments.

Question: Have you ever experienced a full-blown domain crisis in your project? Some misalignment in a core part, very, very deep or fundamental, so big that it was easier/faster to set things aside and rewrite a significant chunk of the system (even whole?) rather than try to gradually improve it?

One thing I wish somebody had told me 10+ years ago is that what you learn about software in training, books, and conferences has as much resemblance to reality as Instagram houses to your life. Yes, there are some useful skills to learn, worthy aspirations to have, etc.

One more reason to keep all my house appliances as dumb as possible. Annoyingly, it's increasingly hard to get "not-smart" versions of everything. Why would I want my treadmill or washing machine to fail without an internet connection? It's more than enough trouble that they need electricity.

Building a house is an endless supply of software inspirations. It never occurred to us to check if windows had a consistent size across walls, though it seems obvious in retrospect. Turns out it matters a lot in windows with glazing bars (local style). If only our architect were Steve Job's fan...

I agree, but for some businesses it's ok to treat devs as just "coders". The understanding is covered by other people or processes. It's like McDonald's vs gourmet restaurant. It's not the question of which is better, they're different businesses. But do you even know which one you are? :)

I love this place. I feel so lucky to have a fully #remotework and can go for a walk to visit the castle and other amazing places nearby. No $$$ can make up for this. Photo by Roman Szymański, our local photographer.

One of the more impactful features I implemented a while ago was matching the order of printed documents with address labels. Programming effort + QA and deployment - 2-3 hrs. The manual effort saved - a few person-hours per week, or over half a month of a full-time salaried position per year.

I'm similar, it's very satisfying when you see how things improve and in the most messy codebases usually you have a lot of quick wins that immediately make a difference. But... that always makes me think of this lady www.instagram.com/reel/C84pfeX...

I think my data-binding library for creating Excel spreadsheets is almost ready for a 1.0 github.com/SimonCropp/E... Already using it in anger at work. Spreadsheet code that would have taken a few hours to build now takes a few minutes.