78% of Your Developers Aren't Happy. That's a Timebomb

by Stephan Schmidt

Happy 🌞 Sunday,

Welcome to my opinionated newsletter. This week’s insights

• 🔥 Why Hunger Now Beats Tenure for Juniors
• 🌫️ Decoherence: A Fresh Theory of Programming
• 📜 Dijkstra Doubted Natural Language Programming in 1978
• 🔄 Why Everyone Is Starting Over Right Now
• 💸 Why LLM Prices Dropped 80% in Months
• 🎯 Getting Good at AI Takes Real Practice
• 🪟 The 33-Year-Old App Still Running on Windows 11
• 🏢 IBM Says Open Weight Models Are Good Enough
• 📸 Rare Photos From Inside NeXT
• 💣 The Developer Unhappiness Timebomb

📅 Reminder: I’ll be at LeadDev Berlin on November 9 and 10, next week! Meeting people today is more important than ever. Use the code AMAZINGCTO for €200 off standard tickets, and say hi if you’re there, for a coffee, a beer - if you have an exciting idea or opinion, I’ll pay for both - or your opportunity to tell me where I’m wrong. Disclosure: AmazingCTO is also a media partner of LeadDev Berlin.

Good reading, have a nice Sunday ❤️ and a great week,

Stephan
CTO-Coach and CTO-veteran

If you only read one thing

Hunger beats tenure: Advice for junior engineers (7 minute read)

“This month, I laid the traditional coding interview to rest. “ At last. The challenges I had with new hires were always motivation, attitude, culture fit, ownership, getting things done - not bad technical skills (we didn’t build rockets). AI has now brought out what was true for a long time, since Google came to the scene or StackOverflow (not copy paste coders), where you didn’t need the deepest technical knowledge for most tasks. “Well, good news: Execution isn’t expensive anymore.” What to do? “Pick a radical idea: Come up with a few ideas that would, if successful, dramatically improve the chances that your organization achieves those goals. Be ambitious!” YES! #RadicalChangeRadicalResults

https://felixrieseberg.com/hunger-beats-tenure-advice-for-junior-engineers/

Stories I’ve enjoyed this week

The Decoherence Trap: When AI slop drowns out the signal in an endless sea of noise (44 minute read)

“This essay introduces the concept of decoherence: the accumulation of locally valid but mutually incompatible or redundant models within a system, such that the cost of change is dominated by translating among those models and preventing unintended side-effects.” (not LLM models)

I have seen many theories about programming, but haven’t seen a good one for a while (the last one was Idea Flow). This is a new, sophisticated theory of programming, of a mutator walking a subgraph of the system (e.g. one class of your codebase) and mutating it. And how the concept of decoherence changes the task, when a decoherent system can be changed faster, when it can be changed slower compared to a coherent one. Not sure I agree with the AI findings, but brilliant nevertheless.

https://maevyn.codes/decoherence

E.W.Dijkstra Archive: On the foolishness of ’natural language programming’. (EWD 667) (6 minute read)

“In order to make machines significantly easier to use, it has been proposed (to try) to design machines that we could instruct in our native tongues. [..] But is the argument valid? I doubt.” Interesting article written in 1978, nearly 50 years ago. We are now in the era of natural language programming (and soon there will be no programming at all). What is still true from the article? What looks ridiculous? Interesting read. “Therefore, although changing to communication between machine and man conducted in the latter’s native tongue would greatly increase the machine’s burden, we have to challenge the assumption that this would simplify man’s life.” ?

https://www.cs.utexas.edu/~EWD/transcriptions/EWD06xx/EWD667.html

Everyone is starting over — Gabriel Valdivia (7 minute read)

“Well, I’ve got good and bad news: we’re all starting over.” Or as I say, it’s the wild wild West and no one knows what they are doing. “Being enthusiastic about change, instead of merely tolerating it, opens up possibilities that didn’t exist even a year ago. [..] David Bowie spent decades as one of the most famous people on the planet and still treated his own identity like something he was allowed to scrap and redo.” (most resistance to AI is an identiy crisis) When everything is changing, there are no rules, yes, everything is possible: “Back on the Zoom call, the question is still sitting there: if I were starting over today, what would I focus on?”

https://www.gabrielvaldivia.com/notes/everybody-is-starting-over

OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? (8 minute read)

“OpenAI announced GPT-6 Sol at at a shocking price of $2 / $10 / $0.20 that halves the input and output price relative to the temporarily-discounted GPT-5.6 Sol” A nice deep dive of prices/costs of OpenAI and Anthropic over the span of 2026. If you want - or need (CFO breathing down your neck) - to understand token pricing and #trends, this article is for you.

https://minimaxir.com/2026/10/cheap-llms/

Getting good at AI takes time and practice (9 minute read)

“Many people think that getting “good at AI” is trivial. “ I see this again and again, here is the AI, then we do a workshop or a hackathon, then off you go! Then I see engineers who use AIs only to generate code, I have CTOs who still ask me what to do when a “key” developer leaves with their knowledge, managers that plan with 3 months onboarding of new hires. Many engineers do not use their AI to find security problems, contradictory domain models (decoherence!) or uneven architecture. They don’t use AI to plan and do massive refactorings or migrate frameworks. Getting good at AI takes time and practice.

https://www.seangoedecke.com/getting-good-at-ai-takes-time-and-practice/

Oldest GUI application that still runs on Windows 11 (3 minute read)

The oldest app the author could find that still runs on Windows 11 is *33 years and 8 months * old. This is the reason Windows was (is?) great for business, and this is the reason why today it looks like a mess. Apple did the opposite, make apps obsolete but move ahead with OS features. Which resulted in many apps I had that don’t run anymore. I see CTO clients of mine stuck with legacy code bases and exposed APIs, they now need to support. The question of how do you force customers to upgrade their systems can be lethal for B2B SaaS companies. Same as Win 11 vs. OS X. I wonder if you support your APIs in 50 years.

https://blog.enlyth.com/posts/oldest-windows-gui-app-win-11/

IBM CEO says open weight models are now irresistibly good and cheap on-prem (4 minute read)

“The CEO said open weight models currently deliver “anywhere from 25 percent to 10 percent of the token cost of the frontier models. That is very attractive given that they’re now 90 percent as effective.”” Do you use open weight models? Does your CEO demand you find out how to do that? First the IBM CEO says this, then there will be a Harvard Business Review article, then it will be - for Germans - in the Manager magazine, then your CEO will ask you, are you prepared? No is not a possible answer I predict.

https://www.theregister.com/systems/2026/10/22/ibm-ceo-says-open-weight-models-are-now-irresistibly-good-and-cheap-on-prem/5304640

NeXT Computer, Inc. | Douglas Menuez Photography Collection - Spotlight Exhibits (512 minute read)

I used a NeXT computer at university, loved the machine, it was so way beyond the SUNs we had. And as a day-to-day working machine also above the DEC Alphas (though I loved their raw power). NeXT. Imagine Steve Jobs, following his vision, getting kicked out of Apple, the company he founded, didn’t give up but founded another company to follow his vision. That’s grit. NeXTs were beautiful in every way.

https://exhibits.stanford.edu/menuez/browse/next-computer-inc

Stack Overflow Developer Survey 2026 (9 minute read)

22% Happy at Work, 33% Not Happy at Work, 45% Complacent at Work - a timebomb that will go off in the near future. I already see more and more resistance, developers digging in. The mathematicians from the last newsletter to call to not use AI. The developer who said artists are fighting AI, so shouldn’t we too? Revolution anyone?

https://survey.stackoverflow.co/2026

What is Stephan doing?

Working on an idea I had yesterday: a fuzzer that fuzzes WASM files and finds problems in them (security, crashes, bugs, n+1). The idea is marketing drops a WASM binary for engineering to run, and engineering uses a fuzzer to find all problems in the binary. Really exciting work on the problem of “How can we let marketing ship to production?”