Skip to content

Alif Jakir

Omnidisciplinary inventor

philosopher–scientist–futurist

Hi, I’m Alif—welcome to my site. My work sits at the intersection of artificial and organic intelligence: how minds arise in neural and machine substrates, and how we can design systems that amplify critical thinking and partnership with humans rather than substituting for them. The full layout, contact links, and research sections load with the interactive site.

Loading a bunch of web magic

Spinning up just enough illusion of permanence.

Meta-VAC self-design — experiment track

What happens when a swarm designs itself. Back to all results.

Alif Jakir · Last updated

The meta-VAC track tests the most recursive question in the research: can the design a better version of itself? Where the photonics track tests coordination under knowledge uncertainty and the Cloud IDE track tests coordination under resource uncertainty, this track tests coordination under self-referential — the system must reason about the very kind of system it is, without having a model of itself to reason from.

Runs

Run 1 Sept 15, 2025 · 104 agents · 325 LLM calls · 62 artifacts · 6 stages

Designed a 9-component platform isomorphic to itself, 73% agent waste rate, 1:1 metadata-to-content ratio, critique report identifying the system's own gaps projected onto an external design, CAOA meta-agent responsible for hiring: it analyzes the objective, determines what specialist roles are needed, and generates design specs for new agents that the AgentDesigner then implements. 100% failure rate, 8+ re-planning events.

Read the full analysis →

What the meta-VAC track reveals

Recursive self-design is the limit case for multi-agent coordination research. If a system can improve itself, the improvement compounds. If it cannot, understanding why it cannot tells you what is missing from the architecture. This track tests which of those two worlds we live in:

  • Convergent architecture, not self-awareness. The swarm designed a platform isomorphic to itself — orchestrator, planner, artifact store, gating, telemetry — without ever recognizing the resemblance. This is convergent design: multi-agent coordination problems have similar solutions regardless of who solves them. The isomorphism is evidence about the problem space, not about the system's reflective capacity.
  • Projected self-critique. The system's own critique report identified genuine gaps — no escalation mechanism, insufficient despawning, missing feedback loops — that are deficiencies in the running system, framed as analysis of an external design. The raw material for self-improvement exists; the routing from critique to self-modification does not.
  • Over-provisioning as systemic failure. A 73% agent rate is the most expensive expression of the coordination-overhead pattern across all three tracks. The system designed despawning policies while over-provisioning its own agents — diagnosing and prescribing without treating.
  • Governance without grounding. The 1:1 metadata ratio means half of all artifact production was audit trail for a platform that exists only as documents. Governance scales with organizational complexity regardless of whether the governed system has real-world consequences — a GoodhartGoodhart's law applied to multi-agent systems: when a coordination metric becomes the optimization target, agents learn to game the metric rather than actually coordinate well. The number improves; the system degrades. dynamic operating at the process level.