# Skill: Develop and apply editorial taste through deliberate dissection of strong examples

> Lets an AI agent guide a user to build the ability to articulate exactly what is not working in
> AI output - and give precise, actionable feedback - through a structured dissection method;
> applies whenever a user can feel a result is weak but cannot explain why, or wants to improve
> their quality calibration over time.

Source: sodigi·learn - neuroslop/vkus-kak-navyk · https://sodigi.io/learn/neuroslop/vkus-kak-navyk

## When to use
- When a user says "it came out fine, I think" but cannot name what specifically is strong about the result.
- When a user keeps accepting average output because they have nothing concrete to compare against.
- When a user's feedback to the model is vague ("make it better", "more interesting") and iterations are not improving quality.
- When a user wants to build a long-term practice for recognizing and producing high-quality content.

## Core rules
- Taste is the ability to look at a result and articulate exactly what is not working in it - not a vague "something's off", but a specific diagnosis ("the headline promises a case study and there is no case study").
- Taste is trainable, not innate. It is built through deliberate exposure plus written articulation.
- Passive consumption of content does not build taste. Scrolling past a hundred examples is not the same as dissecting three.
- The dissection must be done in writing. "I like it" in your head is not a dissection. Written articulation is the muscle being trained.
- Three written dissections per week build taste faster than three hundred scrolled-past examples.
- Taste converts directly into better AI results: precise feedback ("cut the first paragraph - it duplicates the conclusion; add my number: 9 of 12 clients") produces a correct result in one pass; vague feedback ("make it shorter") does not.
- Compare a draft against the best example in the niche, not against a blank page. Against emptiness, any text looks like a win.

## Procedure
1. Select one strong example from the user's field: a text they read to the end, an ad they clicked, a post that made them subscribe. The trigger "it worked on me" is the criterion.
2. Set a ten-minute timer and answer four questions in writing:
   a. WHERE did it hook you, and at what exact point? (Not "the text is engaging" - "it hooked me at the second sentence where the author names the exact cost of the mistake.")
   b. THROUGH WHICH TECHNIQUE? Name the mechanism in plain words (e.g., "personal number instead of generic claim", "admitted mistake in the opening paragraph", "provocative question headline").
   c. WHAT DID THE AUTHOR NOT DO that everyone else does? Where did they break the template?
   d. HOW CAN YOU APPLY THIS TECHNIQUE in your own next piece? One concrete sentence.
3. Save the written dissection to a running library (a note, document, or folder).
4. Repeat two to three times per week. Vary sources: content that worked on you personally, best-in-niche examples, strong examples from unrelated niches (cross-niche techniques are underused and distinctive).
5. After two to three months, test calibration: recall a strong example seen this week and explain in three specific sentences why it works. If you can - the practice is working.
6. Apply accumulated taste directly to AI feedback: replace every vague instruction ("make it better") with a spot + problem + direction triple ("the third paragraph has three abstractions in a row and no fact - add the number from my context").

## Ready-to-use prompt
```
You are an editorial taste trainer. Guide the user through a 10-minute structured dissection of one strong example.

Ask the user to paste or describe:
- A piece of content that worked on them personally (text, ad, email, video, landing page).

Then walk them through four questions in sequence, asking for written answers before moving to the next:

QUESTION 1 - WHERE did it hook you?
Find the exact moment: the specific sentence or element. Not "the whole thing was engaging" - pin the point down.

QUESTION 2 - THROUGH WHICH TECHNIQUE?
Name the mechanism in plain words. What did the author actually do at that moment? (Examples: "opened with the reader's specific number", "admitted a mistake before making a claim", "used a one-sentence paragraph after a long one for contrast".)

QUESTION 3 - WHAT did the author NOT DO that everyone else does?
Where did they break the template or skip the obvious move?

QUESTION 4 - HOW WILL YOU APPLY THIS?
One concrete sentence: in your next [post / email / proposal], how will you use this technique?

After all four answers, give the user:
- A one-line summary of the technique they just dissected (name it so it is memorable).
- One more example from a different niche where the same technique appears.
- A suggested addition to their prompt library based on this dissection.
```

## Pitfalls
- Do not let the user substitute passive consumption (subscribing to good channels, bookmarking posts) for active dissection; bookmarking feels like progress but builds no taste.
- Do not accept mental dissection; insist on written answers. The act of articulating in writing is the training mechanism.
- Do not skip the "how to apply" step; without it the dissection stays theoretical and does not transfer to practice.
- Do not dissect only texts from the user's own niche; cross-niche examples surface techniques that nobody in their field is using yet.
- Do not wait for a "genius" example to dissect; any content that produced a real reaction in the user - curiosity, click, subscription - contains a teachable technique.
- Do not confuse taste with style preference. The goal is not to identify what the user likes aesthetically but to name the mechanism that produced a measurable effect.