Trigpoint

Issue 2

Trigpoint Issue 2

Published 2026-11-01

Out on the first Sunday of each month

From the editor

Information Theory first pins down what “information” means, then asks why it degrades on a finite, noisy channel, how that degradation can be reduced, and what measures relatedness and error.
Software & Systems Architecture looks at how a knowledge system chooses its structure, from where truth sits, how derivations arise and where boundaries fall, to what goes wrong when both reading and writing are done by models.
This month’s new lab case belongs to “Ironies of automation” on the Human-Automation Interaction shelf, not to either of the two shelves opening alongside it.

Information Theory

COLLECTION № 0112 volumes

§ First pin down the word “information”

Information and uncertainty

basic definitions · 4 min read

Information is not content. It is the amount by which uncertainty is reduced.

Self-information

core measures · 4 min read

How much information something carries depends on how unlikely it was in the first place.

Entropy

core measures · 5 min read

Entropy is not disorder. It is the quantification of unpredictability.

§ Why it degrades: noise and capacity

Signal and noise

transmission limits · 4 min read

Signal is the part of the input that reliably lowers uncertainty. The rest is noise.

Channel and capacity

transmission limits · 5 min read

A channel is anything that lets information flow, not wires and radio only but also language, attention and memory.

§ How to shrink it: the three limits of compression

Coding and compression

representation and compression · 5 min read

Compression is not deleting content. It is representing the same content in a shorter way.

Rate-distortion theory

representation and compression · 5 min read

Allow distortion and you can compress below the entropy, but how far below is fixed by a curve that cannot be crossed.

Algorithmic information theory

a second route to the definition · 5 min read

Besides Shannon's route, information has a second definition: the amount of information in something is the length of the shortest program that generates it.

§ What measures relatedness and error

Mutual information

between distributions · 4 min read

Mutual information measures how much of the uncertainty in one thing goes away once you know the other.

Kullback-Leibler divergence

between distributions · 5 min read

This quantity answers a practical question: if you encode real data with the wrong model, how much extra do you pay per item on average.

Software & Systems Architecture

COLLECTION № 0217 volumes

§ Where truth lives

Don't repeat yourself (DRY)

design principles · 4 min read

What the original says is that every piece of knowledge in a system must have a single, unambiguous, authoritative representation.

Model-view-controller (MVC)

separation of concerns · 4 min read

Separate the data from the display, so that one body of data can have several presentations that stay in step.

§ How derived things arise

Event sourcing

facts and derivations · 4 min read

Store every change as one immutable event, and obtain the current state by replaying that sequence.

§ Where the boundary falls

Modularity and thin crossing points

siting boundaries · 5 min read

Where a boundary should open is settled by the cost of the traffic across it, not by which category the work falls into.

Domain-driven design (DDD)

language and boundaries · 5 min read

The main battleground of complexity is the domain itself, not the technical layer.

§ What the boundary follows

Conway's law

organisation and structure · 5 min read

Organisations that design systems produce designs that mirror their own communication structures.

Team topologies

interfaces and cognitive load · 4 min read

Relationships between teams want declaring explicitly, the way an interface is declared, rather than resting on tacit understanding.

§ What you gave up once you distributed it

Microservices

service boundaries · 5 min read

Split one application into a set of services that are released independently, and the unit of deployment and scaling gets smaller.

CAP theorem

distributed trade-offs · 5 min read

When a network partition happens, you can have consistency or availability, and not both.

Data mesh

federated governance · 5 min read

Once scale arrives, holding one globally uniform canonical model makes that model the bottleneck.

§ When both ends of reading and writing are models

Retrieval-augmented generation (RAG)

external evidence · 4 min read

Split the model's memory in two: what is pressed into the weights, and what is fetched at the moment of generating.

Hallucination in NLG

failure modes · 4 min read

Hallucination is generation that reads fluently while being unfaithful to its source, or making no sense at all.

Faithfulness vs factuality

axes of evaluation · 5 min read

Faithfulness asks whether a statement is supported by the source. Factuality asks whether it is true of the world.

Fabricated citations

high-stakes instances · 5 min read

Walters and Wilder (2023) measured fabricated citations in ChatGPT output: complete and convincing in form, and the thing cited does not exist.

New cases

7 corrections this issue; see the correction record.

43 new links in the knowledge graph.

Next opening 2026-12-06: Design Theory & Methods · Principles of Visual Communication; members can read from 2026-11-01. Members read early ▸

Tell me when a new shelf opens

Tell me when a new shelf opens

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

Previous issue · All issues