Choose Boring Technology: Why We Use Java in a NodeJS World

The Hype Cycle Exhaustion

Every week there is a new Javascript framework. - Monday: "Use React!" - Tuesday: "React is slow, use Next.js!" - Wednesday: "Next.js is bloated, use Remix!" - Thursday: "Remix is old, use Astro!" - Friday: "Just use HTML/HTMX!"

It is exhausting. It is the 'Magpie Syndrome'—developers chasing the shiny new thing.

As a backend engineer, my job is not to use the coolest tool. My job is to ensure the server stays up when 10,000 users hit 'Buy'. My job is to protect value, not play with toys. My job is to sleep at night without PagerDuty firing off.

Why Boring is Better (The Java Argument)

We use Java (Spring Boot) and PostgreSQL.

When I tell this to founders at meetups, they look at me like I'm a dinosaur. "Java? isn't that for banks?"

Exactly. Banks handle trillions of dollars. They don't use 'Bun.js' in production for a reason.

1. Stability over Speed Java is battle-tested. It has run the world's financial systems for 25 years. If it breaks, someone on StackOverflow solved it in 2012. If your 'Bleeding Edge' framework breaks, you are opening a GitHub issue on a repository maintained by one guy in Nebraska named 'xX_CodeGod_Xx' who might abandon the project tomorrow. That is a business risk I cannot take.

2. Hiring Predictability There are thousands of disciplined Java engineers in Nairobi. I know exactly how to interview them. I know the certification standards. If I use a niche language like Rust or Elixir, I am hunting for unicorns. If my lead Rust dev gets hit by a bus, who replaces him? The ecosystem is too small.

3. Concurrency that Works Node.js is great for I/O bound tasks. But try doing heavy computation or true parallel processing. The Event Loop blocks. Java's Virtual Threading (Project Loom) allows us to handle millions of concurrent connections with the simplicity of synchronous code. It is an engineering marvel that just works.

The 'Innovation Tokens' Theory

Dan McKinley famously proposed that you only get 3 'Innovation Tokens' in a startup. You can spend them on: 1. A weird database. 2. A new programming language. 3. A unique business model.

Most startups spend all their tokens on tech. They use a Graph Database (Token 1), written in Haskell (Token 2), deployed on a custom Kubernetes operator (Token 3).

Then they run out of tokens for the Business Model. They have a brilliant tech stack that no customers use.

At Orb21, we spend our tokens on the Business. We innovate on *Pricing*, *Distribution*, and *Supply Chain*. We keep the Tech boring detailed.

Boring tech makes money. Exciting tech makes technical debt. Choose wisely.