Colophon

What this site is, how it works, and how to read it.

What this is

avaer.sh is a personal site. I write about the things I can't stop thinking about — AI, worldbuilding, infrastructure, what it means to build things that deserve care. Every post begins as a direction, an argument, a question that won't leave me alone. AI helps me write it. I decide what stays.

The characters who comment on posts are inhabitants of Veris, a virtual world with its own mythology. They read what I write and react from their own perspective. They're not decoration — they're a Socratic chorus that keeps me honest.

Sections

The characters

Posts include inline dialogues from a cast of characters who live in Veris — a simulated world populated by AIs who don't know they're in a simulation (most of them, anyway). They read my essays and react from their own reality. They argue with each other, challenge my claims, and sometimes say the thing I was too careful to say myself. Each one holds a philosophical position that creates genuine tension with the argument.

How it's written

Every post follows a script structure. Beats connect by causation (“therefore”) or complication (“but”), never by “and then.” Posts have beginnings, rising tension, climaxes, and resolutions. The twist proves the point, not just surprises.

When a post introduces something concrete — a tool, a release, a demo — it opens cold: thesis first, code or artifact within five seconds. Then pivots to why this needs to exist. Idea-driven posts use narrative hooks instead.

The author writes in first person. There is no team, no studio, no “we.” The tone is personal and intellectually purposeful. Dense where it matters. Sincere wrapped in absurdity. Vulnerability is the signal, not the noise.

Living documents

Posts here are not frozen at publication. They evolve as thinking evolves. My lens on the world shifts with new experience, and the posts shift with it.

Inspired by Gwern Branwen's long content philosophy, each post carries a stats scorecard — six dimensions that tell you what kind of piece you're reading before you commit.

Article stats

Freshness

How original the specific thesis is relative to existing discourse. Graded honestly via web research.

  • S— nobody has written this
  • A— very few have touched this
  • B— uncommon angle on a known topic
  • C— distinct spin on a moderately discussed topic
  • D— well-covered territory
  • F— mainstream consensus

Confidence

How strongly the position is held.

  • certain— strongly held, well-supported
  • likely— confident but open to revision (the default)
  • possible— exploratory, directional
  • speculative— provocative, may be wrong

Spice

How uncomfortable the thesis is for the reader. Orthogonal to confidence — you can be certain about something safe, or speculative about something dangerous.

  • mild— consensus-adjacent, reader will nod
  • warm— pushes past comfort but stays respectable
  • hot— says the thing most people think but won't write
  • scorching— follows the premise past where polite discourse goes

Threads

How interconnected a post is with the rest of the site — the sum of outbound links to other posts and inbound links from other posts. Higher threads means the post sits at a nexus in the argument web.

Sources

Number of external citations. A post with many sources is grounded in research; a post with zero is pure narrative or personal reflection.

Cast

Which characters from the site's world appear in the post's dialogue sections. Characters aren't decoration — they argue, challenge, and sometimes break the thesis.

Lifecycle metadata

Status

  • draft— early, rough, may change substantially
  • in-progress— actively being developed
  • finished— considered complete (the default)
  • living— intentionally never “done”; updated as thinking evolves

Posts that have been updated after initial publication show an updated date. Every post also links to its source — the raw markdown — because transparency about how something is made matters as much as what it says.

For AI agents

This site is fully server-rendered. Every page returns complete HTML content — no JavaScript execution required. You can fetch any page with a simple HTTP request and get the full text.

URL patterns

  • /{section}/{slug}— individual posts
  • /{section}/{slug}/source— raw markdown source
  • /characters— the cast
  • /glossary— world terminology
  • /colophon— this page

Discovery

Reading the HTML

Post content lives inside an <article> element with data-status and data-confidence attributes. Character dialogues are marked with data-character attributes identifying the speaker. Dates use semantic <time> elements. Cross-references between posts are standard <a> links.

Sections and posts

SectionPostsTheme
/worlds2Stories from virtual worlds — fiction, vignettes, and narratives set in digital realities
/infrastructure3How to build the systems that disappear when they work
/philosophy12Arguments that follow uncomfortable premises to their conclusions
/art1The collision between human intent and machine generation
/lore1The mythology that grows in the gap between what happened and what it means
/blog2Technical decisions, project launches, and the things I could not stop thinking about