CARD · Card record

Simon Willison: AI moves the coding bottleneck to cognitive capacity

Simon Willison argues from his own experience that coding agents can multiply a person's output without multiplying their capacity to understand it; when features become cheap to add, conceptual integrity is easier to lose.

01
THE STORY · The original

Read the explanation and its limits.

  • Originalcreator self-reporthttps://simonwillison.net/2026/Aug/19/conceptual-integrity-and-counting-lines-of-code/

Why it is worth understanding

The lesson makes the AI-era information problem concrete: faster output raises, not lowers, the need to filter isolated changes, preserve context, and keep the whole understandable.

Conditions and limitations

This is a creator self-report, not proof that the experience generalizes to every person or team. The public-update snapshot only verifies that the article exists and says this.

Reading progress is saved in this browser. Save or add a note →

02
THE THREAD · Full timeline

3 events, every row traceable.

A card accumulates records as its story develops: new corroborators, market contrasts, material changes at the original — each dated and linked to its archive day.

Card timeline
  1. RecordThe author published an article arguing that cognitive capacity and conceptual integrity become constraints as coding-agent output rises.
  2. RecordThe author’s public updates returned 15 long-form entries. BitShovel retained titles, links, dates, and author metadata—not summaries or article bodies.
  3. RecordA repeat fetch returned the same 15 metadata entries, with no new or metadata-updated entries, so no duplicate information card was created.

03
SOURCES · Sources and evidence

Check each link's purpose and limits.

Sources and evidence · 1 link

Each link has a purpose and scope. Link counts do not establish independent confirmation; direct support applies only to the statements specified below.

Found an error or a problem? Send feedback ↗