Trigpoint

Trigpoint

Give every claim a reference point to stand on.

AI connects anything, sound or not.

Connected where sourced, blank where not.

Pick a shelf and read ▸ Or go straight to the index ▸

Collections

A theory library where every entry names its original sources.

2 shelves · 23 volumes

October 2026

  1. COLLECTION № 01 Human-Automation Interaction Open the shelf ▸ 11 volumes · 13 lab cases · Every source: bibliography checked ✓
  2. COLLECTION № 02 Personal Knowledge Management Open the shelf ▸ 12 volumes · 13 lab cases · Every source: bibliography checked ✓ “Knowledge management is picking a good tool”: that is a misreading.

Opening soon

  • November 2026Opens 1 November 2026Diffusion & Networks · Situations & Heuristics · Growth Analytics
  • December 2026Opens 6 December 2026Design · Visual Communication
  • January 2027Information Theory · Computer Science
  • February 2027Informal Fallacies · Adversarial Epistemics
  • March 2027Decision Rights & Division of Labor

Tell me when a new shelf opens

Shelf notices are sent only when a new shelf opens or when the reading rules change.

How to use

Three steps to use Trigpoint

№ 01

Pick a shelf and read

Start from the first volume of any shelf. Read it and go.

№ 02

Compare against the Lab cases

First-hand evidence from our own systems, with results recorded as they came, positive or not.

№ 03

Trigpoint Pass for the application layer

The verification layer of every node, full case texts and the doctrine clauses.

Get Trigpoint Pass ▸
Trigpoint’s position: read those who came before, stand on the shoulders of giants, then talk about innovation.

We do not claim zero errors. We claim traceability. Every page states how far its sources were checked.

Our ground rules ▸
Problem space

What you may have been assuming

“Hand it to the AI and I will just keep an eye on it.”
The fault is structural: sustained monitoring is the human's weakest post, and the more reliable the system, the looser the watch.
← AnchorsVigilance decrementAutomation complacency and bias
→ Task-allocation reading card ▸
“If the AI gets it wrong, I will notice.”
The fault is in the assumption: once you are out of the loop your awareness of system state has already degraded, so you are not in the state when the error arrives.
← AnchorsOut-of-the-loop performance problemSituation awareness
→ Takeover reading card ▸
“Whether you trust the AI is a matter of feel.”
The fault is in the selection: trust is a calibratable variable and each mismatch has its own name. Over-trust is misuse; abandonment is disuse.
← AnchorsTrust in automationUse, misuse, disuse, abuse
→ Trust-calibration reading card ▸
“The more automation the better.”
The fault is in the dimension: automation is a continuous scale, and every level you climb trades one failure mode for another.
← AnchorsLevels of automationIronies of automation
→ LOA allocation reading card ▸
“Every note belongs in exactly one place.”
The fault is structural: one place leaves one path, and finding it means drilling down the classification level by level. The remedy is to lay further paths alongside the classification, not to tear the classification down.
← AnchorsMemex
→ Retrieval-path reading card ▸
“A tidy folder tree means the material is organised.”
The fault is in the selection: tidiness is the standard at filing time, and a way of organising is tested by whether you can lay hands on the material when you need it. Sorting by the task at hand, or by what the reader is trying to do right now, never starts from which topic the content belongs to.
← AnchorsPARA methodDiátaxis
→ Organising-scheme reading card ▸
Honesty discipline

How we work

Read the original before the summary. Where the two disagree, the original governs.

Say verified when it is verified and unverified when it is not. Where we do not know, we write that we do not know.

Attribute correctly: the originator of a concept and the coiner of its name are recorded separately.

Cases come only from our own tests. We do not invent incidents to make up numbers.

The theory layer is open to all readers. Payment covers only the verification layer and the case layer.

Our ground rules ▸