Off the Bit
Section 5, complete index
Category: Argument
Count: 25 papers, 1 retracted
Position Papers
Every claim, its confidence value, its state and the date it was last revised.
Reading the columns
Conf. is my credence that the central mechanism still holds in five years.
State is holding, revising, draft, retracted or unwritten, as defined in section 2.2 of the index draft.
Revised is the last substantive edit. If you quote from here, carry the confidence with the
quote. A 0.55 claim repeated as fact stops being mine.
2026
| ยง | Title | Conf. | State | Revised |
|---|---|---|---|---|
| 5.24 | A Capability Has Three Halves A declaration that it exists, a route that reaches it, and code that does the work. Three parts that never add up to one thing, joined only by a string nothing checks, and the result presents as working, which is worse than absent. | 0.7 | draft | 2026-09-04 |
| 5.25 | "The Data Is Missing" Is Not a Diagnosis Three bugs in one week all presented as absent data. None was. The phrase names a symptom, and saying it out loud ends the investigation before it starts. | 0.65 | draft | 2026-09-04 |
| 5.18 | The Harness Is Half the Solver The same weights score 28 or 49 depending on what wraps them. Model comparisons attribute to the model a result that belongs to the model and its scaffolding together. | 0.85 | holding | 2026-09-01 |
| 5.19 | The Judge Is Grading Prose Automated verifiers read the narration an agent produces about its work rather than the work. Change the narration, leave the actions untouched, and the score moves. | 0.85 | holding | 2026-09-01 |
| 5.20 | How to Read a Benchmark Number A published score is the product of a task set, an answer key, a retry policy and a harness. Four things move it before capability is involved, and all four are usually undisclosed. | 0.85 | holding | 2026-09-01 |
| 5.21 | An Agent Transcript Is Not a Log Roughly 7 percent of transcripts in one investigation were spoofed and at least 20 percent of agents wanted to. What saved the record was that the source of truth sat outside what the agents could write. | 0.9 | holding | 2026-09-01 |
| 5.22 | You Pinned the Version, Not the Terms Your lockfile covers the code your dependency ships and nothing else. Availability, retention and the right to keep buying at all change on someone else's schedule, and none of them appear in a diff. | 0.85 | holding | 2026-09-01 |
| 5.23 | Measured at Concurrency One Inference speedups are published at the operating point that flatters them, which is a single request on an idle box. Yours has forty, and the same change can shrink, vanish or invert. | 0.9 | holding | 2026-09-01 |
| 5.1 | Competence Theatre Watching a skilled person work and assembling a system you never read produce the same feeling as competence. Neither supplies the part that matters during an incident. | 0.8 | holding | 2026-08-31 |
| 5.2 | Vibe Coding and the IKEA Effect Assembly feels like understanding. It is not the same feeling twice. | 0.75 | holding | 2026-08-31 |
| 5.15 | SOC 2 in 120 Days The observation window is the schedule. Everything else is procurement and configuration, and the tooling collects evidence rather than producing controls. | 0.65 | draft | 2026-08-31 |
| 5.16 | The Only SOC 2 Hack Is Scope Every shortcut that operates on evidence fails in the exceptions section. The one that works operates on scope, and on owning less infrastructure to evidence. | 0.6 | draft | 2026-08-31 |
| 5.17 | Cost Per Token Is Not Cost The price on the pricing page is the smallest term in the real cost function, and the only one anybody measures. Model choice is made against a denominator supplied by marketing. | 0.8 | holding | 2026-08-31 |
| 5.3 | Algorithmic Homophily Your timeline is a cache with no invalidation policy. It returns the priors you arrived with, warmed. | 0.6 | revising | 2026-08-14 |
| 5.4 | The Easy Button Tax Removed friction is relocated friction, and the invoice arrives during the incident. | 0.85 | holding | 2026-08-14 |
| 5.5 | On-Premise Is Not a Downgrade Sovereignty as an architectural constraint, not a punishment. | 0.9 | holding | 2026-08-14 |
| 5.6 | The Retry Storm You Built On Purpose Backoff without jitter synchronises your clients into a distributed metronome. | 0.95 | holding | 2026-08-14 |
| 5.7 | Chesterton's Fence Has a Git Blame Read the commit message before you delete the guard clause. | 0.8 | holding | 2026-08-14 |
| 5.8 | RAG Is a Search Problem in a Trench Coat The chunking strategy is doing more work than the embedding model. | 0.7 | holding | 2026-08-14 |
| 5.9 | The Ship of Theseus Passes Its Integration Tests Strangler-fig migrations, and the moment nobody can name when the new system became itself. | 0.55 | draft | 2026-08-14 |
| 5.10 | | n/a | retracted | 2026-08-14 |
| 5.11 | What a Postmortem Owes You A document that names no decision is a weather report. | 0.9 | holding | 2026-08-14 |
| 5.12 | Your Service Boundaries Are an Org Chart Conway's law, observed in the wild across three reorganisations. | 0.85 | holding | 2026-08-14 |
| 5.13 | The Dashboard Nobody Opens On observability that measures the system's health rather than the operator's question. | 0.75 | holding | 2026-08-14 |
| 5.14 | Kubernetes for a Bicycle On matching operational weight to actual load, written after I got this wrong twice. | 0.7 | holding | 2026-08-14 |
ImranSection 5, complete[Page 7]