synthesis-writing-pitfalls

name	synthesis-writing-pitfalls
description	Universal bad-writing patterns produced by human authors and editors, independent of AI. Covers cringe (humble-bragging, defensive disclaimers, third-person self-reference in personal writing, pre-emptive originality defense, apologizing-while-bragging), throat-clearing, caveat overload, reader-relationship missteps, sentence-level weakness, cliché reliance, and stilted formality. Use when reviewing or editing prose for craft failures, doing archive scans, self-editing drafts, or training writers and editors. Apply alongside synthesis-content-quality (AI-specific patterns) and synthesis-writing-craft (positive principles).
license	CC0-1.0
depends_on	[]
metadata	{"author":"Rajiv Pant","version":"1.0.0","source_repo":"github.com/synthesisengineering/synthesis-skills","source_type":"public"}

Writing Pitfalls

Universal bad-writing patterns that human authors and editors fall into, independent of AI. This skill is the human-source counterpart to synthesis-content-quality (which catches AI-generation patterns) and the negative complement to synthesis-writing-craft (which carries positive principles).

When to Use This Skill

• Reviewing or editing prose for craft failures
• Auditing an existing archive of writing
• Self-editing a draft before publication
• Training writers or editors
• Performing editorial review when AI involvement is unlikely or irrelevant
• Running a comprehensive scan over older or pre-AI content

Why This Is a Separate Skill

Bad writing has two distinct root-cause classes that need different detection and different fixes:

• AI-generation patterns (covered by synthesis-content-quality): statistical regression to the mean, training-data bias, optimization for blandness. Fix is to inject specificity, voice, and expertise.
• Human-source patterns (covered HERE): social and psychological insecurity, defensive posturing, performance of credibility, laziness, lack of confident voice. Fix is to remove defensiveness, trust the reader, and state the substance plainly.

These overlap in surface symptoms (a humble-brag from a human reads similar to a sycophantic phrase from an AI), but the root cause and the fix differ. Treating them as one category obscures both.

Core Philosophy

The patterns in this catalog are rooted in writer-reader dynamics that have existed for centuries. They predate AI and exist independently of it. The fix in every case is some version of: trust the reader, drop the defense, state the substance plainly.

A pattern is identified by its function (what is the writer doing that weakens the prose?), not by surface form. The same words can be a pitfall in one context and appropriate in another. Calibrate by venue, audience, and intent.

The Pattern Catalog

Each pattern is tagged with its confidence tier: [HIGH] (almost always a pitfall), [MED] (often a pitfall but context-dependent), or [LOW] (a pitfall only in certain registers).

Cringe family

1. Humble-bragging [HIGH] — Self-praise wrapped in modesty: "I'm not very good at speaking but I keep getting invited to keynote..." Fix: state the substance directly or omit the framing entirely. Trust the reader to evaluate.

2. Credential-stuffing [HIGH] — Listing roles, awards, or titles inside the body of personal writing where they don't serve the argument: "As someone who has worked at Programmer, Manager, Regional Director, Vice President levels..." Fix: let the analysis demonstrate authority. The byline carries credentials; the body should not repeat them.

3. Pre-emptive originality defense [HIGH] — Disclaiming claims nobody made: "I do not claim these ideas as original." Fix: omit. Confident writers do not preempt plagiarism charges nobody filed. The disclaimer signals insecurity, not intellectual humility.

4. Third-person self-reference in personal writing [HIGH] — Switching to third person in personal essays or blog posts: "the author has been..." in a piece otherwise written in first person. Fix: stay in first person throughout. Third-person self-reference is appropriate in academic CVs and bio blocks; never in personal essays.

5. Apologizing-while-bragging [MED] — Combining a brag and an apology in one move: "I apologize for the length, but I had to cover all twelve speaking engagements." Fix: cut both halves. Either the substance justifies the length (no apology needed) or it does not (cut the substance).

6. Reverse humble-brag [MED] — Publicly downplaying achievements in a way that highlights them: "I was just lucky to be invited to speak at Davos." Fix: state the fact plainly or omit. Either "I spoke at Davos" or skip it.

Throat-clearing

7. Meta-narration [HIGH] — Telling the reader what the article will do instead of doing it: "In this article I will discuss..." or "First I will examine X, then Y, then Z." Fix: just do it. The reader figures out structure from the structure itself.

8. Long preambles before the actual point [MED] — Multiple paragraphs of context before the lede. Fix: open with the point. Context can come after if needed, or in inline asides.

9. "Before I begin..." openings [HIGH] — Any preamble that delays the start: "Before I begin, let me say a word about..." Fix: cut. The article begins when the reader starts reading.

10. Setup-without-payoff [MED] — Building elaborate framing for an insight that does not justify the buildup. Fix: match setup intensity to insight weight. If the insight is small, the setup should be smaller — or the insight needs to be reworked into something that earns the buildup.

Caveat overload

11. More hedges than claims [HIGH] — Every assertion qualified to the point of dissolution: "It might be that, in some cases, perhaps, X could possibly Y." Fix: state the claim. Add ONE precision qualifier if needed. More than one usually means the claim is not ready to make.

12. Defensive parenthetical clusters [MED] — Multiple parenthetical insertions defending against possible objections inside a single sentence. Fix: address objections in their own sentences if they matter, or trust the reader to raise them.

13. "Of course there are exceptions" without naming any [MED] — Generic acknowledgment of complexity that does no work. Fix: either name the exceptions (specific) or omit (let the simple claim stand and accept minor edge cases).

Reader-relationship missteps

14. Reader-flattering postures [HIGH] — Pre-emptive flattery of the reader: "As intelligent readers like you no doubt know..." or "Anyone smart enough to be reading this..." Fix: omit. Flattery weakens both writer and reader.

15. Reader-condescending postures [HIGH] — Pre-emptive talking-down: "Allow me to explain to those unfamiliar..." or "For those who don't know..." Fix: just explain or assume baseline knowledge appropriate to the venue.

16. "As we all know" and variants [HIGH] — "As we all know," "needless to say," "it goes without saying" (yet you are saying it). Fix: cut. If everyone knows, omit. If not everyone knows, just state it.

Sentence-level weakness

17. Empty intensifiers [MED] — Adverbs that do not add information: very, really, actually, literally (when not literal), basically, essentially. Fix: cut. If "X is true" needs "really" to land, the X needs to be stronger, not the adverb.

18. Passive-voice misuse [MED] — Passive where active is clearer and the actor matters. Fix: use active voice when the actor matters. Passive is appropriate when the actor is unknown, unimportant, or deliberately suppressed (e.g., "mistakes were made" is sometimes the precise frame).

19. Buried lede [HIGH] — The most important point comes after several paragraphs of buildup. Fix: move the lede up. Reader attention is highest at the start.

20. Weak openings [MED] — Opening with "I," "There is," "There are," or a long throat-clearing sentence. Fix: open with something concrete. A specific image, claim, or scene.

Cliché reliance and dead metaphors

21. Cliché as substitute for thought [MED] — Familiar phrases that have stopped meaning anything specific: "at the end of the day," "moving the needle," "low-hanging fruit," "thinking outside the box," "a perfect storm," "moving forward," "circle back," "boil the ocean," "drink from the firehose." Fix: state the underlying claim in original language. Cliché is the writer's brain on autopilot.

This is distinct from the AI-typical "exhausted metaphor" pattern (covered in synthesis-content-quality v4.0 as A3-BT-002) — that pattern catches mechanical recombination of training-set phrases like "navigating the complex landscape of." Human cliché reliance is laziness; AI exhausted metaphor is statistical artifact. Different root causes; both pollute prose.

Stilted formality

22. Register mismatch [MED] — Writing in a register that does not fit the venue: corporate-formal in a personal essay, academic in a popular piece, breezy in a serious analysis. Fix: pick the register that matches audience and venue. Personal essays use first person and contractions. Popular technical writing uses active voice and concrete examples. Read a few exemplars in the target venue if unsure.

Article register in a social media post is the dominant modern variant of this pitfall, and its symptoms are now common enough to be cataloged separately. They include: third-person narration of first-person experience ("the site" / "the audit" / "the fix" where it should be "my site" / "my audit" / "my fix"); formal severity labels imported from source material (CRITICAL / HIGH / MEDIUM in a conversational post); essay structure markers ("first", "second", "the principle", "the takeaway", "in summary"); no first-person pronouns in posts about the author's own work; no closing engagement; sentences over 25 words without compression; over-smoothed prose with no rhythmic variation. The same writer's voice should sound different in articles vs social posts — same person, different mode. See synthesis-content-quality criteria 38-41 for the AI-detectable variants and synthesis-content-distribution "Social register vs article register" for the rationale and the expectation gap that drives the failure.

For deeper examples and fix guidance on each pattern, see references/detailed-pitfalls.md.

Detection Process

Step 1: Scan for High-Confidence Indicators

Check for: pre-emptive originality defenses, third-person self-reference in personal writing, "Before I begin," "As we all know," reader-flattering or reader-condescending postures, buried ledes.

If any are present, the prose almost certainly has a craft failure regardless of who wrote it.

Step 2: Count Medium-Confidence Indicators

• 3 or more present: revision warranted
• 5 or more present: substantial revision warranted

Step 3: Assess by Family

• Multiple cringe-family patterns: the writer is performing credibility instead of demonstrating it
• Multiple throat-clearing patterns: the writer has not yet figured out what they are saying
• Multiple caveat-overload patterns: the writer does not believe their own claims
• Multiple reader-relationship missteps: the writer has not figured out who the reader is

The family pattern often points to the underlying problem more clearly than any single indicator.

Step 4: Calibrate by Venue

A pattern that is a clear pitfall in a personal blog post may be neutral or appropriate in a legal brief, an academic paper, or a corporate memo. Apply the catalog with judgment about register.

Quick-Reference Checklist

High-Risk Phrases to Grep For

• "I do not claim these ideas as original"
• "the author has" (in first-person writing)
• "Before I begin," "Allow me to begin"
• "As we all know," "needless to say," "it goes without saying"
• "In this article I will," "In this post I will"
• "Of course there are exceptions" (without specifics following)
• "As intelligent readers like you," "for those unfamiliar"
• "I'm not very good at X but..."
• "I apologize for the length"
• "very," "really," "actually," "literally," "basically," "essentially" (audit cluster usage)

Cringe Patterns Beyond Phrases

• Title or credential listing inside body text of personal writing
• Switch from first person to third person mid-piece
• Closing parenthetical disclaimers about authorship or originality
• Public modesty about widely-known achievements

Structural Checks

• Lede appears in the first paragraph (not the third)
• No meta-narration of the article's structure
• No "before we begin" preambles
• Hedges and qualifiers are sparing, not constant
• Reader is treated as an adult (neither flattered nor talked down to)
• Register matches the venue
• No clichés substituting for original phrasing

Distinct from AI-Pattern Detection

This skill catches HUMAN-SOURCE failures. For AI-generation patterns (saturated vocabulary clustering, exhausted metaphors as structural filler, the concierge tone, hyperbolic subheadings, the rule of three, mechanical transitions, etc.), use synthesis-content-quality. The two skills together cover the bad-writing landscape from both root-cause directions.

For positive principles (what to DO instead of what to avoid), use synthesis-writing-craft.

Related Skills

• synthesis-content-quality — AI-generation patterns and AI-slop detection
• synthesis-writing-craft — Positive principles drawn from the writing-craft tradition
• synthesis-reader-briefing — Pre-writing audience analysis that prevents many of these pitfalls upstream
• synthesis-article-writing — End-to-end article workflow with quality gates
• synthesis-article-refresh — Refresh and revitalize older articles

---

Part of the synthesis writing craft — the writer writes, the AI assists.