One Interface, Every Protocol

Viral outcry

A short, exasperated post from Dax Raad — "idk how people manage infrastructure anymore…" — touched a nerve. It has been reported that within a day over fifty thousand developers saw the thread and piled on: SST, Pulumi, Ansible, "Just stay on AWS," Python REST scripts, job-security jokes — the replies read like a defeatist who's seen it all before. The broader point landed: everyone recognizes the pain of bespoke CLIs and fractured config formats; what differs is the prescription. Tools abound. Foundations do not.

The root cause

The OpenBindings blog argues that lock‑in is a symptom and fragmentation is the disease. Allegedly, stacking more abstraction layers — Terraform, SST, and friends — doesn't solve the underlying problem; it merely moves it. The author reframes the issue through a developer's lens: programming languages solved this long ago with interfaces, protocols and typeclasses. Why can't services do the same? If you code to a shape rather than an implementation, swapping providers shouldn't be a week‑long migration exercise. Simple. Elegant. Hard to pull off at the network boundary, but tantalizingly familiar.

A common protocol?

There's another itch: discovery. Every website has an index.html — a standard front door. APIs have nothing like that. The blog points out this "index.html problem" and even cites recent IETF work; it has been reported that the conversation around RFCs and conventions is restarting. What if you could ask any service, "Can you store a file?" and get a meaningful, standard answer? That's the pitch: define shared shapes for common operations so providers implement them, not the other way around. It sounds almost revolutionary — or at least the kind of boring infrastructure work that finally lets developers sleep at night.

Sources: openbindings.com, Hacker News