06-reference

writewithai writers block as choice making

2026-05-10·reference·source: Write With AI·by Dickie Bush & Cole Schafer
writing-craftai-as-collaboratorprompt-designdecision-makingsanity-check-voice

How To Overcome Writer's Block By Making Better Choices — Dickie Bush & Cole Schafer

Source: https://writewithai.substack.com/p/how-to-overcome-writers-block-by Received: 2026-05-10 via writewithai@substack.com

⚠️ Sponsorship

The piece itself is editorial thought-leadership, but ends with a PS pushing Ghostbase (their AI writing engine product) with a free 7-day trial CTA, plus footer links to Ship30, Premium Ghostwriting Academy, Typeshare — all owned/operated by Bush & Cole. Read the body as craft writing; treat the prompt at the end as the genuine deliverable; ignore the Ghostbase pitch.

Core argument

Writer's block isn't a knowledge problem, it's a choice-avoidance problem. When you're "stuck," you actually have options in front of you — you just don't want to pick one. The act of refusing to choose IS the wrong answer.

Reframe: writing = choice-making. The job is to:

  1. Build enough competence that your options are good ones.
  2. Then make choices on purpose, with intention.

Their internal cycle when stuck on non-fiction:

"What separates great writers from average ones isn't that they have better ideas. It's that they make sharper choices, more often, with more intention."

Where AI fits (their actual recommendation)

Don't ask AI to write for you — that removes the fun and the point. "When you let it do that, you've stopped writing — you're just publishing."

DO use AI to surface the choices you're avoiding. Specifically:

The prompt (the real artifact)

They give a paste-able prompt that hard-constrains the model to surface choices, not write the draft. Key ingredients:

The full prompt is in the source email if needed verbatim.

Why this is in the vault

Two load-bearing reasons:

  1. Direct match to founder's "AI as second brain, not author" stance. The "use AI to widen options, not make decisions" framing is exactly the discipline the Sanity Check voice work needs — Ray-the-COO surfaces drafts, founder makes the choice. Same shape as the founder/operator split.

  2. The prompt itself is reusable infrastructure. The "options + obvious-default-callout + decision-stays-with-human" pattern is a template worth lifting into the /draft-review and /research-brief skill flows where Ray currently sometimes over-narrows toward a single recommendation. Adding an explicit "here are 3-5 options, here's the autopilot pick, you decide" mode would reduce confirmation-bias risk.

Mapping against Ray Data Co

Action candidates

Related