Programmers muse: what if infrastructure were property-tested?

The hunch

It has been reported that a Lobsters user asked readers about their programming hunches — those half-formed ideas you tuck in the back pocket for later. One poster allegedly proposed a striking thought: treat infrastructure-as-code like property-based testing. Instead of writing exact resource declarations, define the properties a correct deployment must satisfy and let the system search for a configuration that meets them. It’s a neat inversion. Hypothesis for Terraform, anyone?

The hurdles

Reality bites. Cloud resources are stateful and come with side effects; the search space for valid configurations is enormous. It has been reported that similar problem framings show up in FPGA placement and Kubernetes scheduling, which suggests the idea isn’t pure fantasy — but it’s also allegedly “very hard.” There are engineering and economic frictions here you can’t paper over with clever syntax.

Why it matters

If it worked, it could change how we reason about reliability and drift: specify outcomes, not step-by-step recipes. That could reduce brittle configs, speed up recovery, even pave the way for automated tradeoff-solving when resources conflict. Sounds like a dream? Sure — but big ideas start as hunches. Who wouldn’t want to tell their stack what “good” looks like and let the system figure out how to get there?

Next steps

The thread also tossed in logic languages — Prolog, MiniKanren — as possible tools to help express and search those properties. It has been reported that the original poster drew inspiration from FPGA work, so there’s already a breadcrumb trail to follow. Interested engineers: wishful thinking, or a call to action? Who’s going to rubber-duck this into a prototype?

Sources: Lobsters