DIGNALEGI

The reading room · Digna Legi

Slop Readers

A personal relevance score

80–100: high value. 70–79: worth the time. Below 70: below the usual publication threshold.

Evidence-reviewed score based on available publisher text. Evidence is complete enough for the main argument, but it does not prove how well every subclaim is supported.

Scores reflect one reader’s profile, not an objective quality rating. Best is a separate personal selection.

How scoring works →

This brief · about 3 min with detail

Original article ↗

Why read this

Bad AI prose is a failure of communicative intent, not em dashes or other surface tells.

AI brief · Checked against source text

The main idea

The author argues that bad AI prose is not mainly caused by em dashes, formulaic phrasing, or an intrinsic inability to mimic style. It is bad because the system has no substantive thing it wants to tell a particular reader, while training and human criticism often reward surface signals of literariness instead of durable meaning. Humans can fall into the same failure when they read by counting stylistic markers rather than asking what was communicated.

Go a little deeper

Style tells are a weak diagnostic

The essay attacks both pro- and anti-AI style scoring as the same degraded act of reading. Positive graders reward dramatic vocabulary and polished sentence shapes; hostile readers count em dashes and stock constructions. In both cases, attention shifts from what the passage is trying to convey to whether it matches a checklist, so the assessment can miss a substantive but awkward piece and overvalue fluent emptiness.

Goodharting writing destroys the target

The author uses “Goodhearting” to describe optimizing for visible proxies of good prose until the real object disappears. If writers merely avoid known AI tells, they are still letting shallow readers set the objective. The better test is whether the writer has a live claim for a real audience; when that exists, style matters, but it stops being the foundation.

Volition is the scarce ingredient

The essay’s useful distinction is between help with expression and replacement of intent. An LLM can produce acceptable prose when the human supplies what to say, whom it is for, the argument’s structure, and the motivation behind it. But once a writer has specified that much, the remaining question is whether the generated output improves on simply publishing the human’s own thinking.

A case from the article

The brisket page that escapes slop

The author gives Claude a plain catering task and gets copy with prices, smoker details, brisket timing, sauce policy, delivery geography, and a refusal to do chicken. The result is still slightly mannered, but it avoids mystical branding because it contains concrete information a customer can act on. The example shows that specificity can beat empty emotional uplift even when generated by AI.

How the case is made

The case is made through close observation, examples of AI-generated catering copy, criticism of style-checking, and comparisons between AI-assisted and author-driven prose.

Where the idea has limits

The argument does not claim AI assistance always makes prose worthless; it distinguishes empty prompt-to-prose generation from cases where a human supplies the message, audience, structure, and motive.

A question to take away · from Digna Legi

When judging a piece of writing, are you reacting to its stylistic fingerprints or to what it made newly thinkable?

What the original adds

The source includes sharper polemic, several named disputes about AI prose, and a practical distinction between LLM-assisted articulation and outsourcing the actual thinking.

About this brief

AI-written, then separately checked for source support, useful detail and clarity. The author’s claims and our editorial question are kept separate. The original remains the author’s work. How we select and summarise →

Digna legi. Worth reading.