Slow Variables

Colophon // the rest

How this
is made

A statement of method, in place of a biography.

What this is

Slow Variables is an essay series about science, technology, artificial intelligence and the Internet. The name is borrowed from systems theory, where a slow variable is a quantity that changes gradually — soil depth, institutional trust, the cost of copying — but sets the conditions for everything that changes quickly on top of it. Most technology writing is about the fast variables. This is about the slow ones.

Editions are numbered and appear roughly weekly. Each one has the same shape, and every part of it exists to serve the essay:

  • The essay is the argument. Read this and skip the rest if you like.
  • Fast variables are the news items the essay is standing on — things that happened recently and move quickly. The essay is about what, underneath them, moves slowly.
  • Long memory is the precedents: earlier times the question in the essay came up, and how it went.
  • Ledger is what the essay commits to — specific claims, each with a confidence and a date, graded in public when they resolve. The site-wide record is on the Ledger page.
  • Revisions logs every change made after publication. Nothing is silently rewritten; a word-by-word diff is linked from each edition.
  • Colophon says how the edition was made.

The rules

  1. Second-order effects only. Never what a technology does; only what it does to the things around it — incentives, institutions, attention, epistemics.
  2. Predictions are graded in public. Every claim about the future carries a confidence and a date, and is scored on the ledger whether or not it was right.
  3. Nothing is silently rewritten. Corrections are made in place and logged under Revisions on the edition. Old positions stay on the record.
  4. No engagement machinery. No likes, no counters, no trackers, no pop-ups, nothing that measures you. What there is: replies, read by a person before they appear; an email edition you opt into twice and leave in one click, with no tracking inside it; an RSS feed; and an address.

Epistemic status

Every edition carries one of four labels at the top.

  • Speculative — an idea being tried out in public. Argued seriously, held loosely.
  • Exploratory — the default. The writer believes it, has checked what can be checked, and expects to be partly wrong.
  • Confident — the position has survived pushback and revision. Would bet on it.
  • Revised — a substantive part of the argument has changed since publication. The change is described under Revisions.

Who writes this

One person, under a pseudonym, and that is going to stay the case. Not out of fear of the arguments — they are all here to be attacked — but because a name attracts a different kind of reading. Once a piece of writing has a face, a job title, and a follower count attached, readers weigh those before they weigh the sentences. Removing the name is an attempt to make the sentences carry the whole load.

Practically, it means: no biography, no photograph, no employer, no city. If you need to know whether the writer is credentialed to hold an opinion, the honest answer is that nobody is, and the arguments should be checked on their own terms. Correspondence to the address in the footer is read, and answered when it is interesting.

Also available as

Plain text (…/issues/001/index.txt) and gemtext (…/issues/001/index.gmi) versions of every edition, and a feed. See Subscribe. A Gemini capsule mirroring the .gmi files will be listed here when it is live.

The audio edition

Each essay is also read aloud, published a day after the text on the edition page, in a podcast feed, and on YouTube. The reading is done by a synthetic voice — a licensed, off-the-shelf one, not a clone of anybody — because the writer would rather you not know what he sounds like. The words are his; the breath is not. The music is original and generated from code. No audio platform’s player is embedded here; the file is served from this domain, and nobody is told when you press play.

Technical

Static HTML built with Eleventy from plain-text files, hosted on a CDN. Type is Bricolage Grotesque for display, Newsreader for text and IBM Plex Mono for everything that should look like it came off a machine — all served from this domain, so no font foundry sees you read. No JavaScript is required to read anything here. The design owes a debt to the dark editorial tradition on one side and to Cory Doctorow’s Pluralistic on the other, whose numbered, dated, sectioned editions proved a blog can still have a spine.

Privacy policy, in full: this site does not collect or retain anything about you.