Convergence vs Imperative Control

Overview

Traditional programming models rely on imperative execution:

do A → do B → do C

Ved introduces a different paradigm:

Define the desired state. Let the system converge to it.

Imperative Control Model

In imperative orchestration:

• execution order is fixed
• success depends on timing
• failures require retries

Example:

create resource wait configure resource retry if failed

Problems:

• fragile ordering
• race conditions
• drift over time

Convergence Model

Ved expresses systems as:

• desired state (goals)
• transition rules

Execution becomes:

while not goal_satisfied: apply transitions

Formal View

Let:

• S = current state
• G(S) = goal predicate

Then:

Sₙ₊₁ = T(Sₙ) until G(S) = true

Key Differences

Example: Worker Scaling

Imperative

if workers < 5: add worker

Ved

goal workers == 5

transition scale_up: if workers < 5: workers += 1

The system continuously stabilizes.

Convergence Properties

Idempotency

Transitions can be applied repeatedly without breaking state.

Stability

System reaches a fixed point where:

G(S) = true

Self-Healing

If external changes occur:

• goals become unsatisfied
• transitions re-trigger

Challenges

Convergence requires:

• well-defined goals
• non-conflicting transitions
• bounded state evolution

Summary

Ved replaces:

step-by-step execution

with:

continuous stabilization

This aligns control-plane programming with:

• control theory
• distributed reconciliation systems