"It's a best practice." That sentence has shut down more technical debates in my career than any real argument. Not because the person saying it was right. Because nobody dared to push back.
Phora's architecture violates at least three best practices. It's a monolith where microservices "should" be. Simple auth where an OAuth provider "should" be. No Redis, no message queue, no Kubernetes. It works. It's fast. I can maintain it alone.
What spreads isn't the knowledge
Every best practice started as a specific answer to a specific problem — somewhere, sometime, for someone. Netflix moved to microservices because hundreds of teams were working on one product simultaneously. Google wrote the SRE handbook because they operate planet-scale systems.
What the industry adopted: the shape. What it left behind: the reason.
I've watched a four-person startup migrate to Kubernetes with five hundred daily users. They spent six months debugging infrastructure instead of shipping product. Not because they were bad engineers. Because they were preparing for a problem that didn't exist.
This isn't stupidity. It's how cargo cults work. You copy the form because that's the visible part. The context is never visible — so it never gets copied.
Seniority isn't knowing more rules
After 16 years of building software, I don't ask what the best practice is. I ask: does this apply to me?
I've made dozens of "wrong" decisions. Didn't split into services. Didn't write comprehensive tests while core features were still taking shape. Didn't follow the "scale early" mantra — there's nothing to scale for yet.
From earlier projects, I've seen the same pattern: the best decisions didn't follow the most rules. They understood the situation. A two-person team is a different context than a fifty-person one. An early-stage product is a different context than a market leader. Failing to acknowledge this isn't following best practices — it's performing a ritual.
One question worth asking
Next time someone says "it's a best practice" — ask: for whom?
If they can't answer, they're not giving advice. They're exporting someone else's problem — without the context.
There's always a next level.
If you like what you see — whether you're building a product or a team — I'd love to hear about it.