📝 Alex's scrapbook

2025-09-10 what should I do differently, ChatGPT?

Sep 19, 202512 min read

  • From What help do I want from Sasha Chapin re writing etc?
  • I’m not necessarily endorsing much of what ChatGPT has written here, but I wanted to experiment with it as I’m not ready for Sasha’s >$300/hr coaching yet
  • I think generative AI is pretty great for strategy stuff and really dreadful for creative stuff, like when it said “write a post with this title!” it was awful, but the tactics-type stuff is good and useful
  • The below are in reverse-chronological order, that is, heading 1 is the most recent reply

1 - Ultra-Concise Summary

  • Except it’s not concise at all because it turns out GPT5 is WORDY AS FUCK DUDE

so it seems like an  ultra-concise, “what to do next” thing is like “start writing a new substack, channeling what’s alive, see where it takes me, and then do a 2nd pass, a 3rd pass, refining it, cutting out the throat-clearing, then share with some friends, then make a video to go along with it”?

note the attached post by Sasha which is why I say that I want to channel what is currently alive (v.s. e.g. starting with a cynical topic in mind like “how to use anger as part of your creative funnel!”, which sounds rancid and linkedin-brained to me) (this is the pdf of Notes Against Note-Taking Systems)

yep—that skeleton is right. let’s lock it into a tiny, repeatable recipe that’s “alive-first” and anti-LinkedIn-brained.

Do-this-next (ultra-concise)

  1. Channel what’s alive (25–40 min).
    • Sit, notice the hottest thing in your system, and write it straight—no research, no “positioning.” This aligns with Sasha’s “most heart-stopping writing comes from synthesizing the unarticulated in the moment.” 
    • “Hottest” is weird phrasing, Chat
  2. Pass 2 — find the kernel (10 min).
    • Highlight the single sacred claim/outcome you actually believe. Paste it to the top in ≀60 words. (This is Sasha’s “write what’s sacred, assume affinity.”) 
  3. Pass 3 — throat-clearing purge (10 min).
    • Nuke hedges, preambles, and explaining-yourself until the piece starts hot. If P1 still reads like setup, delete it. (His exact note: tons of throat-clearing; start at the kernel.) 
  4. Tiny friend read (10 min).
    • Send to 2 friends with two questions only:
      • “What was the first line that made you lean in?”
      • “What sentence felt like setup I could cut?”
    • (You already proposed multi-day refinement + friend feedback—keep it lightweight.) 
  5. Ship the Substack.
    • Title = the promise. One proof, one practice, one vivid moment. (You wanted to publish only the essential—do that.) 
  6. Make the companion video (6–8 min).
    • Cold-open with the promise, show one concrete bit (demo > tell), single CTA. Keep it fresh, not “advicey.” (Matches your plan to pair posts with videos.) 

2. Re: Sasha’s feedback (throat-clearing)

  • Sasha replied and said that my writing is quite moving but has a tonne of throat-clearing preamble in it
  • I told this to ChatGPT and asked if this changes much of the plan it suggested for me

Short answer: yes—Sasha’s note sharpens the plan. We keep the “new work” sprint, but add one non-negotiable: start at the kernel and delete the throat-clearing. He literally flagged “a ton of throat-clearing,” said your post’s heart was moving but took forever to arrive, and suggested you “write what’s sacred,” assume readers already get you, then cross out the remaining preamble. Also: let the Enneagram-3 anxiety of not managing perception happen—that’s good. 

What changes now

  • Kernel-first rule (60 words): open with the sacred claim or outcome in ≀60 words; no context, no disclaimers. Only after the claim do you give one proof, then the how-to. This directly follows his “sacred + assume affinity + delete preamble” guidance. 
  • Throat-clearing purge (mandatory pass): before publishing, run the checklist below; if the first paragraph still reads like hedging or scene-setting, delete it and re-read. Repeat until the piece starts at heat. 
  • Assume “already liked” points: stop pre-justifying who you are or why you’re allowed to say this. Spend those words on the claim, proof, and practice. 

Next 48 hours (all new work)

1) Monday essay (new)

  • Working title:
  • Kernel opener:
  • Then: one concrete proof from this week → the 3-step mini-system → 2 pitfalls → invite to try it today.

2) Thursday video (new)

  • Title:
  • No intro spiel. Cold-open with the promise

3) Friday micro-piece (new)

  • Prompt:

(To quote the man himself, re: why I’m absolutely repulsed by every ChatGPT suggestion of a topic title/theme):

“Your natural salience filter is a great determinant of what’s most alive to you. If you begin to rely on any other filter, you will increasingly record what seems like it should be interesting according to some preexisting criteria rather than what organically sticks to your mind. This is a tradeoff. It is often not a worthy tradeoff.”

Constraint: 120–180 words. First sentence = the claim. No backstory.

Throat-Clearing Purge — fast checklist

Delete or rewrite anything that:

  • Explains why you’re writing (“I want to talk about
”, “I’ve been thinking a lot about
”)
  • Manages perception (“This might be dumb
”, “I know this won’t apply to everyone
”)
  • Recaps biography or travelogue before the claim
  • Over-hedges (“kind of,” “maybe,” “in some ways”)
  • Meta-talks about process (“I’ll probably write more on this later
”)

Tests:

  1. Paragraph-1 guillotine: remove it. If the piece improves, keep it gone.
  2. First-line swap: can your second paragraph be line 1? Make it so.
  3. Stranger scan: would a stranger know what they get by line 2?

Structure to follow on every new piece (for now)

Claim (≀60 words) → One proof (scene or stat) → How-to (1–3 steps) → One vivid moment → Invite. This implements his “sacred first, assume affinity, cut preamble” note exactly. 

One tight question to send Sasha (optional)

  • “Here are two kernel openers (≀60 words) for my next essay—which one is hotter and what one sentence would you cut?”
  • He asked for specific questions, so this matches his bandwidth. 
  • If you want, I can draft the full Monday essay now using the kernel-first template.

3. What should I do next?

Got it—fresh, not fixes. Here’s a tight 14-day “New Work Sprint” with concrete pieces to make, starting today (Wed 10 Sept 2025, Europe/London). Zero refactors. All net-new.

14-Day New Work Sprint (2 essays, 2 videos, 2 micro-pieces)

Week 1 (Sept 10–16)

1) New essay for Monday

Title: (deleting suggestions from Chat because they’re deeply rancid)

Promise to a stranger:

Outline (1,100–1,400 words):

  1. Cold open (2–3 sentences):
  2. The myth:
  3. The system:
  4. Proof of concept:
  5. Pitfalls & fixes:
  6. Invite:

2) Companion video for Thursday

  • Title: (deleting suggestions from Chat because they’re deeply rancid)
  • “Beat sheet”

3) Friday micro-piece (Substack or Notes)

  • Oh shit, I should post my vignettes to Substack Notes!
  • Constraint: ≀180 words, one image or none.
  • Prompt: “Describe the exact moment your day actually starts.” (Coffee smell, keyboard clack, first lie your brain offers.)
  • Goal: texture + voice. No advice.

Week 2

4) New essay for Monday

  • Title:
  • Promise:
  • Outline (900–1,200 words):

5) Companion video for Thursday

  • Title:
  • Beat sheet:

6) Friday micro-piece

  • Constraint: a single, polished paragraph (~120 words).
  • Prompt: “The street where you remember how to be brave.”

4. ChatGPT on my What help do I want from Sasha Chapin re writing etc? post + snippets of both our writings

I read your “What help do I want from Sasha
” page plus your two Substack posts from Ship It Week, and I skimmed a bunch of Sasha’s pieces you linked. Here’s the short of it, then a concrete plan.

What I’m seeing (super quick read)

  • You’ve already proven flow + output: 40 YouTube videos in ~3 months, 400+ subs, a 7.7k-view video with a lively comment thread; you enjoy being on camera. 
  • You want to make YouTube + Substack (+ maybe songs), have ~6 months runway, and you’re asking Sasha for targeted, opinionated guidance. 
  • Your current Substack cadence leans “expressive mode” (alive, first-drafty, friend-coded), and you’ve already intuited the need to publish only the “essential” and write more for non-friends. 
  • You have clear themes (trauma → agency → kenshƍ; creative practice; modern life upgrades) and a good self-diagnosis: title-last, topic-drift, low cohesion. 
  • Sasha’s writing patterns you admire: strong POV posts that reframe (“10x happiness
”, “Notes against
”), vivid “type of guy” portraits, and practical metaphysics (void intimacy). These are useful north stars.

Your next 30 days (do this first)

  1. Pick a simple weekly cadence you can actually keep for 6 weeks
    • Mon: 1 “essential” Substack essay (reader-first; 900–1,500 words).
    • Thu: 1 YouTube anchor (6–8 min) that embodies the essay (not a read-through).
    • Fri: 1 vignette (short, poetic, totally for you)—but only if the Mon essay shipped.
  • This blends your joy (vignettes/video) with audience gravity (one useful essay/week). Your own doc points to “consistency” as Phase 2—let’s actually operationalize it. 

  1. Adopt the “Essential Only” filter you proposed

    • Before you draft: ==write the promise sentence== (what a stranger gets).
    • If you can’t write a crisp promise, it’s not essential yet.
    • You already guessed this move; I’m just making it non-negotiable. 

  2. Title-first, outline-first

    • Your problem isn’t ideas; it’s cohesion. Start with a compelling, specific title and a 5-bullet outline that delivers your promise. You named this as a likely fix—lock it in. 

  3. Refactor one existing post into a reader-magnet

    • Take “Becoming angry” (currently expressive + scattered) and turn it into:
    • New title: “How I Turned Restless Anger into Creative Fuel (A 10-Minute Protocol)”
      • As always, ChatGPT writes the most hideous LinkedIn slop when it comes to “creativity” - but the object-level advice is good
    • Outline: a) why anger ≠ enemy, b) the 10-minute body drill you actually used this week, c) how it changed your writing session, d) pitfalls, e) one invitation.
    • Keep one or two raw paragraphs (your voice) as texture.
    • This respects the alive core while delivering value to non-friends, which you’ve already flagged as the gap.

  4. Relaunch YouTube as the “hub,” but smaller

    • One weekly video tied to the Monday essay. Cold-open with the outcome (“By the end you’ll have X”), then story, then 1 concrete practice. You already planned a hub return—ship it with constraints.
    • đŸ€”đŸ€”đŸ€” What about music too đŸ€”đŸ€”đŸ€”
      • Maybe it’s a written vignette, which is then turned into a song 😎

  5. Distribution ritual (30 minutes/day, timer on)

    • Thoughtful comments on 2–3 adjacent Substacks or videos (add an angle, not a promo).
    • 1–2 tweets/notes that condense Monday’s idea into something quotable.
    • Ask 2 people privately, “If you liked this, who else should see it?”
    • This is the boring “Holy Work” consistency that Sasha says helps 7s stop chasing novelty and actually build something. 

60–90 day arc (how this becomes momentum)

Pillars (rotate weekly):

  • This is the themes that ChatGPT has pulled out from the few writings I shared. Would actually be useful to be like “here’s all my writings, and this fuckin website too, what are my key themes”
  1. Agency & inner tech (how you got from social anxiety → camera ease; void/meditation tactics).
  2. Working-class → weirdly good life** (practical systems; money, friendship, place). Tie to your family arc without getting trapped in memoir. 
  3. Creative practice lab (vignette craft, 3-step editing heuristic, shipping muscles).
  4. Experiments in public (Ship-It-style challenges you run for yourself + readers). 
  • KPIs (weekly):
    • feels like it’s kinda chatting shit here
    • 1 essential post;
    • 1 video;
    • 1,000+ words drafted;
    • 2 external comments/day;
    • 5 email replies solicited (“What part helped most?”).
    • Monetization later—first grow trust.
  • Monetization sniff test (month 3):
    • this feels icky to think about right now
    • consider a light “supporter” tier only if you’ve hit 6–8 consecutive essential posts and you’re getting unsolicited “this helped me do X” replies (Sasha’s “10x happiness” point: more happiness → more capacity; translate that to reader outcomes before paywalls). 

Make Sasha’s life easy (send him 6 laser-specific questions)

  • Not listening to any of this - all gross af
  • He said he’ll answer specifics. Copy/paste a tight note like this:
  1. “Of your post archetypes, which 2 would you double-down on early for Substack growth—and why? (e.g., reframes like ‘10x happiness’, ‘Notes Against
’, or vivid ‘type-of-guy’ portraits?)”
  2. “Looking at this outline for ‘How I Turned Restless Anger into Creative Fuel’, what would you cut/add to make it essential?” 
  3. “From 0→5k and 5k→25k subs, what were your 1–2 most effective distribution habits (not hacks)?” 
  4. “How do you personally decide when a draft is ‘essential enough’ to publish?” 
  5. “Any red-flags in my proposed weekly cadence (1 essay, 1 video, 1 vignette) you’d change?”

This respects his boundaries and invites short, opinionated answers. 

Two quick craft checklists you can use today

The “Essential Essay” checklist (print this):

  • Title makes a promise to a stranger. 
  • 1–2 core ideas, not 7.
  • Concrete practice or takeaway.
  • 1 vivid personal moment for texture (not the whole thing). 
  • Cut 20% of sentences; front-load the value.
  • End with “If you try this, tell me what changed.”

The “7-proof workflow” (to stay out of novelty chasing):

  • 5 minutes of “void sit” → outline → draft → break → edit → ship. 
  • Novelty (music/tweets/extra ideas) after ship.

One note on big life changes

Sasha’s “blow it up” essay is pro-change, but also notes the ~6-month stabilization window. You’re already on “Death Ground” with 6 months of runway—so for this window, keep your environment stable and blow up your process (cadence, editing, distribution) instead. Then reassess at Day 90. 

If you want a fast start this week

  • Refactor “Becoming angry” into the protocol piece and publish Monday. 
  • Record a 6–8 min companion video Thursday demonstrating the protocol (show, don’t tell). 
  • Draft next Monday’s “essential” now, title-first.
  • Send the 6-question note to Sasha with your outline attached. 

You’ve already got the spark and the engine. This plan just puts bumpers on the lane so the ball keeps knocking pins over—week after week.

Graph View

Backlinks