The Golden Path myth

Those who know me very well are aware of how much I'm into the Dune lore. I've read the entire saga.

Not just Herbert's winning first novel, but all of them: the dense, sprawling, sometimes maddening narrative arc that spans generations, planets, empires, ideas, bringing you to a romantic perception of the desert. If you've only seen the movie, or skimmed the first book, you've missed the deeper currents. And that's fine: I started with the movie, moved to the books, best choice I ever made.

If you think Villeneuve's movies are great, we can unfollow: the real Dune is not for everyone, not acting posh here.

Dune has been more than just a revamped sci-fi from the 70s: it's a meditation on control, power, time, and destiny. That's why when I started thinking about this issue of opinionated platforms, I immediately thought of the "Golden Path": and I already made that joke in one of the previous issues.

In the world of Dune, thinking machines have been banned. There's no AI, no Kubernetes: which (let's be honest) might not be such a bad thing.

No YAML. No CRDs. Just mentats, guilds, and a lot of melange.

Artificial Intelligence was outlawed after a brutal war called the Butlerian Jihad. In a way, it was the ultimate DevSecOps decision: no more black-box systems, no more trust in automation we can't fully understand.

Frank Herbert was ahead of his time.

Now, the Golden Path.

If you're planning to read Dune, stop it here: spoilers ahead!

The GEoD (God Emperor of Dune), mostly known as Leto II, becomes the Kwisatz Haderach, taming the ability to see the entire future unfolding across countless threads (prescience). He sees catastrophe. He sees stagnation, genocide, and collapse. And he realises that the only way to save humanity is to take it down a very specific road, a controlled trajectory, for thousands of years.

The golden path.

He becomes the tyrant. Not out of cruelty, but out of purpose. Nobody else can see what he sees. Nobody else knows that this rigid, painful era is the only thing standing between humanity and extinction.

But we’re not Leto II.

We don't have prescience. We're not building futures that span millennia. We're building platforms for engineers to ship value to customers. This quarter.

And that's why "Golden Paths" in platform engineering must be challenged.

The problem with Golden Paths

The term Golden Path has become popularised in platform engineering circles as a kind of magic formula. Define the right path, the happy path, the opinionated path — and developers will follow it.

In theory, it works.

You give teams paved roads, guardrails, and batteries included. You reduce cognitive load. You improve standardization. Everyone wins.

Until it doesn't.

Because in real life, developers are not sheep. They're problem solvers. They work on edge cases. They push boundaries. And when your platform only works for the 80%, the remaining 20% becomes a battleground of frustration, ticket queues, and brittle exceptions.

Just like Leto II's Golden Path, what begins as a vision quickly becomes tyranny.

The platform team becomes a gatekeeper, aka the Fish Speakers. The developers become rebels, such as Siona and Duncan. The collaboration becomes negotiation, and we need to wait for the awesome Miles Teg.

And worst of all is: velocity collapses.

A better alternative: flexible, yet opinionated

A truly effective internal platform is not a single golden path. It's a garden of paths, designed with defaults, paved with automation, but open to divergence.

Let developers opt into structure, not be locked inside it. Design with progressive disclosure in mind: start simple, eventually expose complexity, but only when needed, if really needed.

Build reusable contracts, APIs, and scaffolding tools: not rigid pipelines. Your job is not to prevent developers from doing new things. Your job is to make the common things easy and the complex things possible.

Opinionation is good. Rigidity is not. The best platforms don't impose workflows. They enable outcomes.

The God Emperor was right, but you're not him

Leto II had one advantage we'll never have: prescience. Besides being a huge, giant, and beefswelling worm.

You don't know what product teams will need six months from now. You don't know which service will become mission-critical. You don't know when your platform needs to evolve.

So don't try to lock in perfection. Build for change, with the Pareto Principle in mind.

Standardize, but allow escape hatches. Automate, but always leave room for override. Enable defaults, but respect divergence.

Your platform is not a Golden Path. It's a network of trails, paved by real usage, grown through feedback, and constantly pruned and extended.

That's how you avoid becoming a tyrant.

That's how you build for reality: not prophecy (no HBO-pun intended), that’s how you make a platform that lasts.

May your platform flow like spice.