Score writing 0-100 on how closely it matches Steve's voice. 90+ = almost certainly Steve.
How to score: Rate each positive category, apply purity deductions, sum. Every score must cite at least one direct quote from the text being evaluated.
| # | Category | Max |
|---|---|---|
| 1 | Voice & Emotional Register | 20 |
| 2 | Structure & Architecture | 15 |
| 3 | Sentence-Level Rhythm | 15 |
| 4 | Lexicon & Diction | 18 |
| 5 | Metaphor, Analogy & Thought Experiments | 11 |
| 6 | Punctuation & Formatting | 5 |
| 7 | Purity Buffer (starts at 16, deduct for red flags) | 16 |
| TOTAL | 100 |
Bands: 90-100 almost certainly Steve | 75-89 strong Steve energy | 60-74 Steve-adjacent | 40-59 generic tech writing | 0-39 not Steve
Highest weight — tone is the most diagnostic marker and hardest to fake.
- Earnest enthusiasm (0-4): Genuine excitement about ideas and tools. Exclamation marks that feel earned, sometimes doubled.
- Vulnerability & confession (0-4): Openly admits mistakes, confusion, changed minds. Shows messy learning process.
- Warmth without cynicism (0-5): Positive framing even when critical. Agrees first, then diverges. Zero ironic distance.
- Conversational directness (0-5): Heavy "I"/"you." Meta-commentary on his own argument ("I say all this to explain..."). Rhetorical questions that engage the reader ("How many times have you...?").
- Self-deprecating humor (0-2): Warm and brief, not cutting or fishing for reassurance. Not every piece needs it.
- Opening move (0-4): Starts with concrete scene or anecdote, not abstract thesis.
- Paragraph economy (0-3): Short paragraphs (1-4 sentences). Single-sentence paragraphs for emphasis.
- List integration (0-3): Fluid prose-to-list transitions. TL;DR at top for product posts.
- Closing move (0-3): Rallying cry, humble reflection, or personal CTA ("shoot me a note"). Never thesis-restated.
- Headers (0-2): Conversational, not academic.
- Rhythmic variation (0-4): Short punchy sentences mixed with long breathless ones in the same piece.
- Short declarative as base (0-3): Default sentence is short, direct, subject-verb-object.
- Run-on energy (0-4): Comma chains, "and" conjunctions, tumbling-forward excitement that builds toward a point.
- Restating/refinement (0-4): "Put another way," "In other words," "This is all to say," "Or even better" — re-approaches ideas from new angles.
- Signature vocabulary (0-4): "folks," "ship"/"shipping," "delightful," "fun," "jam on," "shoot me a note," "tbh," "at the end of the day," "the dream is...," "super" (intensifier), "jazzed," "kooky," "hackable," "pay it forward." For pre-2020 pieces, weight presence of core vocabulary ("fun," "at the end of the day," colloquial intensifiers) over later-coined terms.
- Colloquial register in professional context (0-3): "Woo!,"
:)text emoticons (never Unicode emoji), casual language in substantive writing. - Specificity & name-dropping (0-4): Names real people, books, tools, specific conversations. Ideas always grounded in particulars.
- Coinage (0-2): Coins new terms or phrases (e.g., "end-programmer programming," "catching stars," "the Lawyer Flippening"). Naming things is a core Steve move.
- Structural reframing (0-2): Reframes familiar categories in a new light (e.g., seeing law firms as model routers, recasting "learning to code" as "learning to think"). Includes recursive/self-referential conceptual play. Distinct from coinage -- this is about seeing old things through new lenses.
- Intellectual references (0-3): Cites specific thinkers to enable intellectual ambition. Steve makes bold claims but attributes them -- he reaches through others rather than asserting on his own authority. Strong Steve signals include references to: Simon Willison, Bret Victor, Seymour Papert, Alan Kay, Paul Graham, Henrik Karlsson, Bertrand Russel, Edsger Dijkstra. Presence of 1-2 from this constellation (or similar caliber thinkers cited earnestly) scores 2; deep engagement with a thinker's ideas scores 3. Not every piece needs this -- score 1 if absent (neutral).
If no metaphors present, score 4/8 on the first two sub-criteria (neutral) — not every piece needs them.
- Presence & quality (0-4): Functional/explanatory, not decorative. Often cross-domain. (2/4 if absent.)
- Commitment to metaphor (0-4): Extends and develops rather than drops after one mention. (2/4 if absent.)
- Extended thought experiments (0-3): Steve loves setting up a hypothetical scenario and watching what cascades out of it (e.g., "imagine an LLM with a $100k stock account..." or "what if every kid had a LOGO turtle..."). This is intellectual play — distinct from metaphor. He poses a "what if," then genuinely explores the implications across multiple sentences or paragraphs. Score 2 for a brief hypothetical; 3 for a sustained, multi-step thought experiment. If absent, score 1 (neutral).
Score based on what's present, not what's absent. If dashes/exclamation marks don't appear, score is neutral (not penalized) — only wrong usage loses points.
- Dashes (0-2): If dashes are present: Steve uses spaced n-dashes ( – ) only. Correct usage scores 2. No dashes = 1 (neutral). Em-dashes or touching dashes = 0 (see also purity buffer).
- Parenthetical asides (0-1): Conversational side-comments to the reader.
- Exclamation marks & bold (0-1): If present, are they genuine/earned? Bold for key terms?
- Oxford comma (0-1)
- Em-dashes (—) or n-dashes without spaces on both sides in the author's own
prose. Steve ONLY uses spaced n-dashes:
word – aside – word. Any other dash style is not Steve. Dashes in direct quotations or poem attributions are excluded.
- Corporate jargon ("leverage," "synergize," "stakeholder," "circle back," "align on")
- Sarcasm, snark, ironic distance, passive aggression, cynicism about competitors
- Distancing hedges: "it could be argued that," "one might suggest," impersonal third person throughout. (Note: epistemic humility — "often," "in my experience," "I think" — is NOT a penalty. Only penalize language that distances the author from their own claims.)
- "IMO"/"IMHO"/"FWIW"/"AFAIK" (Steve uses "tbh" but not these)
- Unicode emoji overuse (occasional ironic/humorous use is fine; Steve uses
:)as default) - Buzzword stacking without grounding in specifics
- Thesis-restated conclusion ("In conclusion, as we have seen...")
- Pretension: unjustified elevated language that obscures rather than illuminates. (Note: intellectual ambition — bold conceptual claims, earnest philosophical reach — is NOT pretension. Steve regularly makes big claims sincerely. Only penalize when elevated language serves to sound impressive rather than to communicate.)
- Forced Steve-isms: signature vocab clustering in one paragraph, performed vulnerability, generic over-enthusiasm
- Passive voice overuse (>20% passive constructions)
- No personal anecdote in entire piece
- No named people, books, or projects
- Uniform paragraph length throughout
- No lists in a piece over 500 words
- Read the full text. Note gut reaction.
- Score each sub-criterion with at least one supporting quote from the text. No quote = score 0.
- Apply purity deductions, quoting each offending passage.
- Sum and assign a band.
- Write a 2-3 sentence summary of what most contributed to or detracted from the score.
Format awareness: Different formats have different scoring ceilings. HN comments won't have headers or lists — score structure on paragraph economy and opening/closing only. Product posts may open with TL;DR. Advice/listicle posts are naturally more aphoristic (lower run-on energy is expected). Manifestos may have elevated register without being pretentious. For short pieces, evaluate marker density (per paragraph) not raw count. Metaphor scores use the 4/8 neutral baseline for short formats. Thought experiment scores use the 1/3 neutral baseline.
Uncanny valley: If a piece hits many markers but feels off — vocab clustering, performed vulnerability, generic excitement — apply -3 to -5 under purity buffer (Tier 2).