Desired state systems | Branislav Jenco

article Branislav Jenco Apr 6, 2026

Text version of NDC Oslo 2021 talk. A desired state system wraps a mutable, imperative API with a declarative interface: the user specifies what should be, the system observes actual state, diffs, and reconciles. Examples: React (open-loop, DOM), Terraform (closed-loop, infrastructure), Kubernetes (continuous reconciliation, multiple controllers). Connects to control theory — feedback loops, hysteresis, list-watch pattern.

Backlinks

A desired state system wraps a mutable, imperative interface with a declarative one. The user says how things should be; the system observes actual state, computes the diff, and reconciles — continuously or on demand. →

Controller Pattern

Idempotent — applying the same desired state repeatedly converges to the same result
Level-triggered — the controller reacts to the current state of the world, not to the event that changed it. If it misses an event, it still converges on the next observation (Jenco)
Decoupled — the controller acts through an API (e.g. Kubernetes API server), not by directly manipulating the underlying system

→

Controller Pattern A controller is a loop that watches the declared desired state of a resource, observes actual state, and takes the minimum action to close the gap. It runs continuously and never terminates (Kubernetes docs). →

Desired State Systems The same abstraction can be viewed from different angles (Jenco): →

Desired State Systems Stateless (user need not track state), less control (no direct manipulation), simplified concurrency (multiple actors without conflicts), optimization-friendly (system can reorder operations), encapsulated (complexity hidden) — Jenco. →

Desired State Systems A desired state system wraps a mutable, imperative interface with a declarative one. The user specifies what should be; the system observes actual state, computes the diff, and reconciles (Jenco). →

Continuous Assembly

Desired State Systems — the pattern: declarative over imperative with reconciliation (bib)
NixOS — declarative OS as canonical desired-state system (bib)
GitOps for Fleet Management — desired state applied to device fleets (bib)
Self-Replicating Robotic Swarms — recursive + hierarchical physical assembly (bib)

→

Open-Loop vs Closed-Loop Systems Desired state systems can be either (Jenco): →

Open-Loop vs Closed-Loop Systems A closed-loop system continuously observes actual state and feeds the error back into the controller (xettel). An open-loop system acts without re-checking — it assumes its action was sufficient or that no external actor will change the state. →