This is fantastic.
02.03.2026 15:10
👍 32
🔁 2
💬 0
📌 0
This is fantastic.
Whenever you feel anxiety about the technology updating increasingly fast, remember that Haas F1 web page runs on PHP + Drupal, and Ferrari's is on Nextjs 9 😅
My guess would be vibe coded projects with AI (based on old training data) and general more downloads due to overall more projects? (thanks to many people trying out new projects with AI)
Photo of Power Loom at Manchester Museum of Science and Indsutry
Photo of description of Power loom Power loom T. Larmuth & Co., Manchester, around 1860 By the early 19th century, new machines like this power loom could make cloth more quickly and cheaply than people. Groups of angry handloom weavers raided cotton mills at night. They burnt and broke power looms to protest against the new technology. How do you think you would react to technology replacing your skilled work?
Photo of ribbon loom with Jacquard head at Manchester Museum of Science and Industry
Text of description of Ribbon Loom with Jacquard head: Ribbon loom with Jacquard head T. F. Wilkinson Ltd, Coventry, 1900 At the top of this loom is a programming system. Cards with small, punched holes programmed the loom to weave different designs. It made woven patterned cloth much easier and cheaper to produce. The punch card system inspired the invention of the earliest computer. Weavers using Jacquard looms, 1900. Past Pix / Science & Society Picture Library
Power looms and Jacquard looms at the Manchester Science Museum today.
Automation of skilled work, protests against new technology...
"How do you think you would react to technology replacing your skilled work?"
IMHO, perfect match between domain, user vocabulary and code will probably never exist. But every creation and realization of wrongness is a knowledge gain and a step in the right direction. Asking questions and discussion them is for sure also valuable. (And 👍 for the open question style)
Definitely - I think building the model from code is the most used way for devs but also the hardest. I hope some of AI gains is that we get more used to communicating style (although of course AI communication are not as valuable as conversations with real users)
Like talking to real users? Talking to previous devs of the project? Appreciating the insights of our dedicated test team? ...
NEW POST
Fragments: the future of senior developers, junior developers, more on cognitive debt, DevEx versus AgentEx, the role of IDEs, consequences of task switching in supervisory programming
martinfowler.com/fragments/20...
NEW POST
Powerful context engineering is becoming a huge part of the developer experience of modern LLM tools. @birgitta410.bsky.social explains the current state of context configuration features, using Claude Code as an example.
martinfowler.com/articles/exp...
Great, great talk about Design and UX (although the title is a little bit misleading) www.youtube.com/watch?v=1fZT...
This is why ISO certificates mean absolutely nothing
PostNL has ISO certificates on everything... but then ships this kind of bug to production, making it unusable. Clearly no E2E tests, no monitoring
(We always knew ISO was BS for software development - but now there's proof)
Fragments: Thoughtworks launches AI/works™, electricity consumption of LLMs, how LLMs shift where we put discipline, and thoughts on The Situation in Minnesota
martinfowler.com/fragments/20...
I have learned that about two-thirds of all software projects are neither delivered on time nor within budget. At first, this seemed ridiculous to me. It doesn't anymore, despite having worked with many smart people for ten years since. Join my thoughts in my new blog 🙏 dev.to/shaman-appre...
![The "AI is going to replace devs" hype is over – 22-year dev veteran Jason Lengstorf [Podcast #201]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:chiyruymvbjhtllotxdxzacr/bafkreib7sgtys4t64lra7hoqs5myyfpienq4fxa3kluh3l2quya7elkk3i)
A great podcast that focuses, among other things, on the human side of the AI hype while also acknowledging its benefits. youtu.be/lIghF_OewYg?...
ah, that makes perfectly sense. Yes, I don't use SSR.
I strictly separate frontend (Angular) code and backend (NestJS) code. I put shared code into a plain (npm) module, and install it in frontend and backend relative from file path like "../shared" or use npm workspaces. If you have more advanced use cases maybe utilize something like nx.
The ability to write Angular libraries which are not hard coupled to an Angular version, so that I don't have to upgrade all my libraries with each major Angular version.
Maybe this is helpful to you: I use this simple self written middleware in NestJS for that: github.com/shaman-appre...
I have put together a (long overdue!) draft RFD on using LLMs at @oxide.computer, but I know that there is a ton more to be said on the topic; thoughts and experiences welcome!
rfd.shared.oxide.computer/rfd/0576
100% agree. prompting -> reading -> accepting is a start, but sitting down and actual working with it gives you a much deeper understanding.
To be honest, I am not even sure, if right now the secondary costs exceed the gains of our genie - although, I would miss my genie as tool for sure
I absolutely feel the pain and call it "secondary cost of AI" - the time we invest in learning, configuring and controlling the genie, that we could have spent elsewhere.
Ich wünschte, es wäre eine sehr gute, überspitze Parodie; aber eigentlich meiner Meinung nach eine sehr gute (und witzige) Zusammenfassung des Status quos
www.youtube.com/watch?v=Bj3m...
As a general positive side-effect, which no code approaches didn't had: Writing specs for the AI (aka reasoning about the specs and writing docu) has now somehow become "sexy" for everyone.
They can't make it work - They haven't burned to much time. Now, the are hopefully more sensible about the trade-offs of invested time / possible customization etc, as they have felt some pain themselves.
Even if the (AI-) code will need expensive maintenance in the future, it will be a net profit for sure; in the worst case it can be re-developed (by an expensive developer) from scratch again with a validated user workflow already in place for reference.
They can make it work - great! They saved hand-offs, communication overhead, got immediate feedback...
Now, none-developers cannot only "expect both", but quickly try it themselves with different outputs:
But for some reason, the business experts almost always require all the customization options that are only available with a 3GL. This is not a judgement. It is just an observation from more than 30 years in that business and I see little evidence that this is going to change. Thus, my cautious prediction is that AI solutions will suffer the same fate as 4GL, MDA, No code/Low code and alike because business experts (as well as the responsible decision makers) expect to get both: Concise requirements descriptions in a few sentences while having all degrees of freedom of a 3GL at hand – which cannot work by definition.
Great, great series about Software Development in general by @ufried.bsky.social. I wish more people would talk and think about this; especially with AI in mind. www.ufried.com/blog/softwar...
The only point where my view differs 👇
