Jonas Notes

open-loops

3 min read

Open Loops

Open loops are unresolved questions already present in SWAP or the current AI/ notes.

Active AI-System Loops

Personal R&D frontier state model

Source: Pathfinder - Proposed Curiosity Agent Stack, Pathfinder - Curiosity Stack Implementation

  • What is the minimum useful state model for a personal curiosity system?
  • How often should notes be distilled before recommendations become repetitive?
  • What should count as “recently explored”?
  • How should the system avoid turning every weak idea into a backlog item?

Main chat versus worker boundary

Source: Pathfinder Orchestration Architecture, Bastion - Cursor as Primary Runtime Plan

  • Which tasks should remain in the main chat?
  • Which tasks should be delegated to worker sessions?
  • What context package is enough for a worker without bloating the main session?
  • When does ACP-first become insufficient and require a first-class provider integration?

Agent observability MVP

Source: OpenClaw Observer VM for Agent Observability

  • What is the smallest event schema that makes agent workflow review useful?
  • Should the observer start from session digests, structured events, or sampled transcripts?
  • Which scorecard dimensions matter most: clarity, missing context, wasted tokens, delegation quality, or human prompt quality?
  • How much logging is useful before privacy and retention become the main problem?

Backend and Systems Loops

Serverless Postgres and connection coupling

Source: SWAP

  • Why is PostgreSQL perceived as tightly coupled in serverless environments?
  • What specific problem does Neon solve?
  • Is the main issue connection management, storage/compute separation, cold starts, transaction semantics, or something else?

Kafka and event-driven systems

Source: SWAP

  • What makes Kafka different from a queue, log, or pub/sub broker?
  • When does Kafka become unnecessary operational weight?
  • How should a deep dive connect storage internals, consumer groups, ordering, compaction, and exactly-once claims?

OLAP, OLTP, and storage engines

Source: SWAP, ClickHouse MCP for Dataservice Trace Analysis

  • What are the real storage and query-shape differences between OLAP and OLTP systems?
  • How does ClickHouse fit into that distinction?
  • What should a practical storage-engine note explain for system design use?

Kubernetes networking and sidecars

Source: SWAP

  • How do VPCs and Kubernetes networking actually interact?
  • When are sidecars the right abstraction versus unnecessary complexity?
  • How do service meshes change the operational model?

Spring ecosystem internals

Source: SWAP

  • What is the difference between Spring Boot and Spring Cloud?
  • How do Tomcat and Spring Boot request chains fit together?
  • What JVM and GC details matter for backend engineering practice?

Buildable Project Loops

Personal spending ingestion

Source: SWAP

  • Can enough transactions be captured without reading arbitrary iOS notifications?
  • Is email alert parsing plus monthly statement reconciliation sufficient?
  • How should Apple Pay Shortcuts data be treated when it is timely but incomplete?
  • What reconciliation algorithm best handles reimbursements, PayNow repayments, missing repayments, and reward categories?

Obsidian graph-neighborhood plugin

Source: SWAP

  • Can a plugin highlight a node and its neighbors like graph-view hover behavior?
  • Is the right implementation a graph-view extension, a canvas-like view, or a separate local visualization?

Chat-history LLM bot

Source: SWAP

  • What chat history is available and exportable?
  • What should be treated as memory versus private source data?
  • Should this be a retrieval bot, a summarizer, or a personality/context simulator?

Business and Finance Loops

Value of being informed

Source: SWAP

  • Does it pay to be informed?
  • In which domains does information advantage translate into action, and where does it become entertainment?

Market and company structure

Source: SWAP

  • Why are there many stock exchanges?
  • Why does IPO enable expansion for tech companies?
  • What are the two exit paths of a tech firm?
  • Why did Apple, iPhone, Facebook, and TikTok succeed?

Personal finance decisions

Source: SWAP

  • Where should emergency funds be stored?
  • How should credit card reward constraints affect spending and card selection?

Backlinks