the worst architecture meeting is the one that starts with the technology. “Let's go microservices, Kafka, and that database from last week's blog post.” No one asked what business problem it solves. Six months later, the trending stack has become the anchor stalling the very growth it was supposed to support.
Choosing technology isn't picking a sports team. It's a business decision with consequences that last years. And the sober choice almost always beats the excited one.
§ 01 / ThesisThe best stack is the one that fits your problem.
There's no best stack in the abstract. There's the one that serves your operation, your team, and the growth you project — and that you can maintain when the vendor leaves the room. The right technology is the one your team can sustain and that holds up under growth, not the one that wins a LinkedIn argument. For most mid-to-large companies, a well-built monolith goes further than a constellation of microservices no one can operate.
§ 02 / The hidden costComplexity you adopt too early.
Every piece that enters the stack carries a cost of maintenance, of hiring, and of mental load. Microservices solve a scale problem you may never have — and charge you, from day one, network latency, deploy complexity, and the need for senior people to operate it.
- Adopt complexity when the pain arrives, not in anticipation of it. Scaling later is a good problem to have; paying for scale that never came is guaranteed waste.
- Count the cost to operate, not just to build. The question isn't “can we do it with this?” It's “who maintains this a year from now?”
Org structure leaks into architecture. A small team that picks a big-tech stack builds its own trap.
§ 03 / No lock-inA reversible bet beats an irreversible one.
A good stack decision is one you can undo without rewriting everything. Clear boundaries, isolated dependencies, data that exports in an open format. It's not about avoiding vendors — it's about not being held hostage by one. Nothing that enters the architecture should require the original vendor to keep running.
§ 04 / ClosingTailored is about your body, not the shop window.
A tailored stack doesn't mean an exotic one. It means fitted to your business, your team, and your horizon — just as a tailored suit is about your body, not what's in the window. The right choice is almost always less exciting and far more durable.
Before choosing the technology, finish the business conversation. The stack is a consequence, not a starting point.
end · field note #51 · noûs / apr 26