H
Howardism
Howardism · Vol. 03Plate I · No. 01

Essay, filed.

Pieces60SectionEssayOldest10 Apr 2026Newest1 Oct 2026

Longer-form arguments and reflections.

  1. S01
    Adaptive Probing, Not Standardization: Splitting the Information Side of the AI-Interview Effect
    · 11′
  2. S02
    What the Agent-PR Oversight Numbers Can and Cannot Say
    DerivedCode ReviewAI Coding Workflow+2
    · 14′
  3. S03
    Is Self-GC's 0.3 Commit Threshold Portable? Re-Deriving the Cache-Break Break-Even
    · 9′
  4. S04
    Competitor-Indexed Safety Triggers: What the Revision Record Settles About Pause Postures, Acceleration Regret and Benchmark Perimeters
    DerivedGovernanceAI Policy+2
    · 15′
  5. S05
    Construct or Roster? What BEI's Entanglement Graph and IRT's Ability Ranking Are Measuring
    · 18′
  6. S06
    The Fable Gate: Who It Caps, and What in the Safety Case Is Independent of the Model
    DerivedSafetySafeguards+1
    · 11′
  7. S07
    Harness Patterns Under Scale and Domain Shift: Context Routing, Other Domains, Large Action Spaces, and the Overseer
    · 18′
  8. S08
    Rationale as a Dated Record: Where the Why Lives for the Next Reader
    DerivedAI Coding WorkflowKnowledge Management+2
    · 16′
  9. S09
    Harness Value Is a Product, Not a Score — Why the Artifact-Payoff Questions Keep Returning Partially Answered
    DerivedHarnessAgent Engineering+2
    · 14′
  10. S10
    Does the Augmentation/Automation Split Govern Skill at Work?
    DerivedWorkforceHuman Capital+2
    · 22′
  11. S11
    Memory-Poisoning Numbers, Conditioned on the Write
    DerivedSecurityMemory+1
    · 26′
  12. S12
    Is Persistence the Line Between Prompting and Spec-Driven Development?
    DerivedAI Coding WorkflowPlanning+1
    · 19′
  13. S13
    Writer/Reviewer vs Agent-to-Agent Review
    DerivedCode ReviewAI Coding Workflow+2
    · 27′
  14. S14
    The Configuration Gap: What an Assurance Argument Measures vs What It Covers
    · 24′
  15. S15
    The Price of Mixing Agents, and the Principal Nobody Counted
    · 28′
  16. S16
    Safety Commitments That Cannot Bind the Actor Who States Them
    DerivedGovernanceAI Policy+2
    · 20′
  17. S17
    The Data Wall and the Validation Commons Are One Supply Constraint
    DerivedData WallVerification+3
    · 22′
  18. S18
    Governance by Benchmark Threshold: What an Index Must Prove Before an Obligation Can Rest on It
    · 28′
  19. S19
    Guarantees That Degrade at Deployment: Action-Space Soundness, Admissibility Without Effect, and a Vendor-Coupled Security Framework
    DerivedAgent EngineeringSecurity+3
    · 34′
  20. S20
    Auditing the Misalignment-Measurement Instruments
    · 27′
  21. S21
    What the Instrument Can Resolve: Two Headline Numbers and Their Missing Denominators
    · 28′
  22. S22
    Authority and Audit Survive Abundance
    DerivedHarnessAuthorization+2
    · 12′
  23. S23
    What Makes a Self-Improvement Artifact Transfer?
    DerivedAgent EngineeringHarness+2
    · 14′
  24. S24
    App Server vs MCP, and the Claude-Side Equivalent: Three Boundaries for Driving Agents
    DerivedProtocolMCP+2
    · 7′
  25. S25
    Bind, Don't Forbid; Prevent, Don't Detect: The Action-Open and Poisoned-Memory Residuals
    DerivedSecurityPrompt Injection+2
    · 10′
  26. S26
    Classifier Gates vs OS Sandboxing: The Defense-in-Depth Story for Auto Mode and Cowork
    DerivedSecurityClaude Code+2
    · 14′
  27. S27
    When Knowledge Layers Disagree: Context Files vs Memory, and Conflicting Sources at Compile Time
    DerivedKnowledge ManagementAgent Engineering+2
    · 7′
  28. S28
    The HTML Artifact Lifecycle: Where Plan History Lives, and When Disposable Becomes Durable
    DerivedHuman AI CollaborationAgent Engineering+2
    · 6′
  29. S29
    Is Human Review of AI-Authored Code Still a Real Control, or Already Rubber-Stamping?
    DerivedAI Coding WorkflowCode Review+2
    · 17′
  30. S30
    Is Breadth Cheap Now? Specialist Ramp Speed and Domain-Expert-as-Builder at Scale
    DerivedWorkforceExpertise+2
    · 6′
  31. S31
    Does 'Impossible, Not Tedious' Kill Defense-in-Depth? Layered Friction, Agent-Relativity, and the Frequency Paradox
    DerivedSecurityZero Trust+2
    · 13′
  32. S32
    The Orchestrator's Real Workload: Decision Burden, Framing Discipline, and Whether Taste Scales
    DerivedFounderOversight+2
    · 8′
  33. S33
    Oversight When the Signals Give Out: the Activation Fallback and the Taste Reward
    DerivedAlignmentMonitoring+2
    · 14′
  34. S34
    Playbook Boundary Conditions: the Devil's-Advocate Substrate and the Prototype's Edge
    DerivedProduct ProcessFounder+2
    · 8′
  35. S35
    Verifying Without a Compiler: Cowork's Harness vs Claude Code's, and Why the Slice Verifier Stays
    DerivedVerificationCowork+2
    · 6′
  36. S36
    What Scaffolding Survives Model Improvement — and How Do You Know When a Line Turns Harmful?
    DerivedHarnessAgent Engineering+2
    · 9′
  37. S37
    How Much Signal Do Public Benchmarks Still Carry — and What Replaces Them?
    · 14′
  38. S38
    Can Models Learn to Separate Instructions from Data? Durable Property vs Training Gap
    DerivedSecurityPrompt Injection+2
    · 11′
  39. S39
    The Under-Review Divergence: Faros's Widening Crisis vs. CMU's Convergence
    · 9′
  40. S40
    Single General Agent vs. Multi-Agent Coding Architecture
    · 9′
  41. S41
    RSI Growth Curves: Which Friction Binds First?
    DerivedGovernance WorkforceRecursive Self Improvement+3
    · 11′
  42. S42
    The PRD-Replacement Spectrum at AI-Native Speed
    DerivedProduct OrgAI Native Org+3
    · 12′
  43. S43
    Where Does the Why Live?
    DerivedProduct OrgAI Native Org+3
    · 7′
  44. S44
    Foundation → Enterprise → Advanced: Is the Agent Access-Control Jump a Cliff?
    DerivedSecurityZero Trust+2
    · 8′
  45. S45
    How Do You Write Evals for Taste? Character as the Limit Case
    DerivedEvalsLLM Character+3
    · 8′
  46. S46
    Agent Control Plane Patterns: Tickets, Loops, Specs, and Memory Files
    DerivedAgent EngineeringOrchestration+1
    · 9′
  47. S47
    AI-Native Moats Under Frontier-Model Improvement
    DerivedStartupMoats+2
    · 8′
  48. S48
    AI-Native Product Org Bottlenecks
    DerivedProduct OrgAI Native Org+3
    · 9′
  49. S49
    How AI-Native Startups Avoid Speed Becoming Strategic Debt
    DerivedStartupFounder+2
    · 9′
  50. S50
    Where Does Agent Harness Work Remain Durable as Models Improve?
    HarnessAgent EngineeringVerification+1
    · 9′
  51. S51
    The Future of Agent Interfaces
    DerivedAgent EngineeringInterface+1
    · 11′
  52. S52
    Human-in-the-Loop Boundaries
    DerivedHuman AI CollaborationVerification+2
    · 9′
  53. S53
    When Does Verification Quality Determine Whether AI Automation Works?
    DerivedVerificationAgent Engineering+2
    · 11′
  54. S54
    Does the Human-Facing Harness (HTML Artifacts) Hit Its Own Bloat Ceiling?
    HarnessHuman AI CollaborationAgent Engineering+1
    · 6′
  55. S55
    Orchestration vs Employee Framing: Reconciling the Founder's Playbook with HBR's Accountability Evidence
    DerivedFounderAccountability+2
    · 14′
  56. S56
    Opinions on Using AI Tools & the Future of the Software Engineering Role
    CareerAI CoworkingOpinions+2
    · 12′
  57. S57
    Learning to Co-Work with AI: A Software Engineer's Field Guide
    CareerLearning GuideAI Coworking+1
    · 17′
  58. S58
    Opus 4.6 → 4.7 Changes and Multi-Agent Coding Considerations
    · 9′
  59. S59
    When to Use Claude Opus 4.6 for Work
    · 8′
  60. S60
    What Are AI Tools?
    · 3′