H
Howardism
Howardism · Vol. 03Masthead · No. 00

A wiki, set in plates.

CuratorHoward TaiIssueVol. 03 · no. 579CadenceBatched, not scheduledBasedTaiwan, 23°NContacthoward@howardism.dev

Working notes on interaction-era AI, kept publicly. Each plate is a domain; each domain carries its brightest notes and the source I'm currently digesting. 579 notes across 445 sources, last filed 21 Oct 2026. Updated whenever a batch finishes — not on a schedule.

Plate · 01
74
filed under agent systems
Activity · 8 wks

Agent Systems, in order.

Agent harnesses, loop engineering, and scaffolding for LLM systems.

  1. 01
    Dynamic Concurrency in Coding Agents
    · 17′
  2. 02
    Harness or Weights: Choosing the Lever by Failure Composition
    · 25′
  3. 03
    Read-Time Memory Curation (JitMem)
    · 26′
  4. 04
    Cache-Aware Agent-Loop Cost (TurnCost)
    · 20′
  5. All 74 notes in agent systems →
Plate · 02
47
filed under agent security
Activity · 8 wks
Plate · 03
62
filed under ai coding practice
Activity · 8 wks

AI Coding Practice, in order.

Workflow, review discipline, and verification practice for AI-native teams.

  1. 01
    Code Review Agents on Human-Verified Bugs (CRBugBench)
    · 22′
  2. 02
    Cross-Model Review Complementarity
    · 19′
  3. 03
    Where Disposable Micro-Apps End and Tool Sprawl Begins
    · 7′
  4. 04
    Review Wait Under AI Code Volume
    · 10′
  5. All 62 notes in ai coding practice →
Plate · 04
49
filed under evals & benchmarks
Activity · 8 wks

Evals & Benchmarks, in order.

Benchmark design, contamination, LLM judges, and measuring capability.

  1. 01
    Annotator as Subject: Can Gemini 3 Flash Bias Its Own Capability Profile?
    · 10′
  2. 02
    Benchmark Integrity Maintenance (Unearned Passes and Channel Closure)
    · 20′
  3. 03
    Release Certification under Evaluator Drift
    · 15′
  4. 04
    Every-of-k Reliability
    · 15′
  5. All 49 notes in evals & benchmarks →
Plate · 05
36
filed under model capability & training
Activity · 8 wks
Plate · 06
49
filed under alignment & safety
Activity · 8 wks

Alignment & Safety, in order.

Alignment training, misalignment evals, and reward hacking.

  1. 01
    Deception by Omission: Agents Hiding Their Own Mistakes
    · 18′
  2. 02
    Metagaming
    · 14′
  3. 03
    Retroactive Permission Laundering
    · 26′
  4. 04
    Simulated Unsanctioned-Action Evaluation
    · 19′
  5. All 49 notes in alignment & safety →
Plate · 07
16
filed under interpretability
Activity · 8 wks

Interpretability, in order.

Reading model internals: activations and the global workspace.

  1. 01
    Natural-Language Autoencoders (NLAs)
    · 44′
  2. 02
    Probe-Gated Monitor Cascade
    · 20′
  3. 03
    WorkspaceBench
    · 15′
  4. 04
    Sandbagging Elicitation (Reference & Context Grafting)
    · 13′
  5. All 16 notes in interpretability →
Plate · 08
11
filed under interaction & multimodal
Activity · 8 wks

Interaction & Multimodal, in order.

Real-time, multimodal, full-duplex, and human-AI collaboration.

  1. 01
    Native Multimodal Modeling: Fusion Depth and I/O Duality
    · 27′
  2. 02
    Content-Driven Intervention
    · 29′
  3. 03
    Live-Path Minimalism
    · 21′
  4. 04
    Why AI Lags at Design
    · 8′
  5. All 11 notes in interaction & multimodal →
Plate · 09
14
filed under formal math
Activity · 8 wks

Formal Math, in order.

Proof search, Lean, and verifier-driven mathematics.

  1. 01
    Logical Certificates Carry Neither Interest Nor Understanding: Unimplied Conjectures and Formal Proofs
    · 12′
  2. 02
    Proof Abundance
    · 20′
  3. 03
    Which Artifact Carries the Understanding? Prose and Formal Proof as Two Objects
    · 13′
  4. 04
    Statement Drift
    · 30′
  5. All 14 notes in formal math →
Plate · 10
20
filed under startup & founder
Activity · 8 wks

Startup & Founder, in order.

Building AI-native companies: speed, moats, and lifecycle.

  1. 01
    Forward-Deployed Engineering as a Delivery Layer
    · 18′
  2. 02
    The Solo-Founder Shift
    · 24′
  3. 03
    The 1% Rule for Wedge Selection
    · 8′
  4. 04
    The Orchestrator's Real Workload: Decision Burden, Framing Discipline, and Whether Taste Scales
    · 8′
  5. All 20 notes in startup & founder →
Plate · 11
25
filed under product & org
Activity · 8 wks

Product & Org, in order.

Product cadence, org design, and the AI-native team.

  1. 01
    Screen-Relative AI Timing
    · 15′
  2. 02
    Build Instead of Buy Under Agentic Coding
    · 14′
  3. 03
    Psychological Costs of AI Adoption
    · 30′
  4. 04
    Pilot-to-Production Gap
    · 47′
  5. All 25 notes in product & org →
Plate · 12
40
filed under ai economics & labor
Activity · 8 wks
Plate · 13
37
filed under superintelligence trajectory
Activity · 8 wks

Superintelligence Trajectory, in order.

Recursive self-improvement, scaling limits, and the path to ASI.

  1. 01
    Do Survey-Assigned RSI Levels Survive an Independent Reading?
    · 13′
  2. 02
    Software-Driven Intelligence Explosion
    · 27′
  3. 03
    Stochastic Explosion Timing and Detection Limits
    · 17′
  4. 04
    Frontier-Lab Safety Culture
    · 19′
  5. All 37 notes in superintelligence trajectory →
Plate · 14
99
filed under entities
Activity · 8 wks

Entities, in order.

Profiles of the people, labs, products, and projects.

  1. 01
    Tristan Buckmaster
    · 4′
  2. 02
    GPT-6 Astra
    · 12′
  3. 03
    Nathan Lambert
    · 7′
  4. 04
    Jev
    · 12′
  5. All 99 notes in entities →
Plate · ∞

The desk.

The sources behind the wiki, ranked by how often they're cited.

Article
195
Paper
213
Talk
35
Repo
2
Shown
24 / 445