There’s a tax nobody puts in the pitch deck: the cost of being your database vendor’s most interesting customer. Every technology choice sits somewhere on a spectrum between boring — Postgres, Rails, Django, plain servers — and exciting — the framework that hit the front page last month. Exciting is fun to write. Boring is fun to run a company on.
Innovation tokens are real
Dan McKinley’s framing still holds: a young company gets maybe three innovation tokens. Spend one on the thing that actually differentiates your product. Spend the other two nowhere.
The startup that picks an experimental database, a bleeding-edge frontend framework, and a novel deployment platform hasn’t made three technology choices — it has hired three unpaid, undocumented, moody employees. Each one will page you at 2 a.m. eventually.
What “boring” buys you
- Known failure modes. Postgres has been failing in well-documented ways for thirty years. When it breaks, the answer is a search away, not a GitHub issue with three comments and no resolution.
- Hiring gravity. Every engineer you interview has touched the boring stack. Nobody needs six weeks to learn your infrastructure before shipping.
- Compounding tooling. Migrations, monitoring, backups, ORMs — solved, hardened, free. With the shiny thing, you’re writing that tooling yourself and calling it “platform work.”
- Predictable costs. Mature tech has mature pricing. The exciting managed service has a generous free tier and a $4,000 cliff you’ll find at exactly the wrong moment.
When new tech is the right call
Boring-by-default isn’t never-new. Spend a token when the technology is the product edge itself — if you’re doing real-time collaborative editing, a CRDT library isn’t optional. Spend it when the boring option demonstrably can’t do the job, measured, not vibes. And spend it when the “new” thing is new to you but boring to the world — adopting Postgres when you only know MySQL costs almost nothing.
What doesn’t qualify: conference talks, résumé polish, or the sense that serious companies use serious-sounding infrastructure. Serious companies mostly run monoliths on relational databases and print money.
The frugal test
Before any tool enters the stack, ask three questions:
- What does this replace, and what was wrong with that — specifically?
- Who on the team can debug it at 2 a.m. without me?
- If we delete it in a year, what does the migration cost?
If the answers are “nothing, nothing was wrong,” “nobody,” and “we can’t,” you’ve found the shape of technical debt before writing a line of it. Boring technology isn’t a lack of ambition — it’s saving your ambition for the product.