Non-Obvious Pick vs Top-Ranked Shortlist in ADHD: What’s the Real Difference?

The non-obvious pick highlights a single, novel-yet-viable idea selected for maximum creativity, while the top-ranked shortlist provides a configurable set of the highest-scoring solutions ranked by overall merit, with both outputs generated simultaneously by the critic module in UditAkhourii/adhd.

The UditAkhourii/adhd repository implements a multi-phase ideation system that produces two distinct result categories after the divergence phase. While both outputs originate from the same critic evaluation pipeline, they differ fundamentally in selection criteria, quantity, and strategic purpose.

What Is the Non-Obvious Pick?

The non-obvious pick represents a single idea flagged specifically for its novelty and surprise value. After the divergence phase generates candidate solutions, the critic scores each idea across three dimensions: novelty, viability, and fit. The system surfaces the idea with the highest novelty score—the most surprising yet still workable solution—and marks it with the visual indicator ★ non-obvious pick →.

This selection appears as a single line in the rendered output, prefixed with the star marker. For example:

★ non-obvious pick → rage-quit = instant abort + branch to cheaper/faster model

The rendering logic that adds this marker resides in src/render.ts at line 53. Because the selection prioritizes novelty over raw composite scores, this pick often surfaces creative solutions that baseline approaches might overlook.

What Is the Top-Ranked Shortlist?

The top-ranked shortlist is a curated list containing the highest-scoring ideas overall, regardless of whether they represent conventional or unexpected solutions. The same critic that evaluates novelty produces a numeric ranking for each idea based on the combined score of novelty, viability, and fit. The system returns the top K results (controlled by the topK configuration flag) as the shortlist.

Unlike the non-obvious pick, the shortlist may contain obvious textbook solutions alongside innovative ones. In the rendered output, this appears after the non-obvious pick as part of the result.shortlist array. As documented in the repository’s README at lines 64-66, the output format follows this pattern:

shortlist: scout-fork …; daemonize …; race 3 LLM replicas

The shortlist serves as the comprehensive "best of" collection, providing multiple viable options ranked strictly by total score.

Key Differences Between Non-Obvious Pick and Top-Ranked Shortlist

Understanding the distinction requires examining three core differentiators: selection logic, output quantity, and display purpose.

Selection Criteria: Novelty vs Combined Score

The non-obvious pick uses novelty as its primary driver. The critic identifies the single idea with the highest novelty rating that still meets viability thresholds. In contrast, the shortlist ranks items by the overall combined score (novelty + viability + fit). This means a highly practical but conventional solution might rank first in the shortlist while never qualifying as the non-obvious pick.

Quantity: Single Item vs Multiple Items

There is exactly one non-obvious pick per run, or none if no idea meets the novelty threshold. The shortlist, however, contains multiple items—typically defaulting to a configurable topK count. This quantitative difference reflects their distinct roles: the former highlights a specific creative direction, while the latter provides a menu of validated options.

Display Purpose: Highlight vs Shortlist

The non-obvious pick functions as a highlighted recommendation designed to draw attention toward unexpected, innovative solutions. The shortlist functions as a ranked catalog of the strongest candidates overall. Users should consult the non-obvious pick when seeking breakthrough ideas, and the shortlist when evaluating the full spectrum of high-quality options.

Technical Implementation in the Source

Both outputs originate from the same critic evaluation pass, diverging only in how the results are filtered and rendered. In src/render.ts at line 53, the code checks for the non-obvious selection and prepends the star marker before emitting the final output. The shortlist is populated by sorting all scored ideas and slicing the array according to the topK parameter, then appending this list to result.shortlist.

The README example at lines 64-66 demonstrates the sequential display: the starred non-obvious pick appears first, followed immediately by the semicolon-separated shortlist entries.

Summary

• The non-obvious pick is a single, novelty-maximized idea marked with ★, selected for creative value rather than total score.
• The top-ranked shortlist contains the highest overall scoring ideas (quantity controlled by topK), including conventional solutions.
• Both use the same critic scoring pass but filter results differently: novelty-only versus combined-score ranking.
• Rendering occurs in src/render.ts, with the non-obvious pick receiving special prefix treatment before the standard shortlist display.

Frequently Asked Questions

Can the non-obvious pick also appear in the top-ranked shortlist?

Yes. While the non-obvious pick is selected via a novelty-specific filter, it often scores highly enough on overall metrics to appear within the result.shortlist array. The shortlist may contain the non-obvious idea alongside other high-performing candidates, including obvious textbook solutions.

How do I configure the number of items in the top-ranked shortlist?

The shortlist length is controlled by the topK flag. Adjusting this parameter changes how many of the highest-scoring ideas the system returns after the critic evaluation completes, though it does not affect the single-item non-obvious pick selection.

What happens if no idea meets the novelty threshold for a non-obvious pick?

The system returns exactly one non-obvious pick only when an idea satisfies the novelty criteria. If all generated ideas fall below the novelty threshold or fail viability checks, this output category will be omitted, and only the standard top-ranked shortlist will appear in the results.

Is the non-obvious pick always the highest-scoring idea overall?

Not necessarily. Because the non-obvious pick prioritizes the novelty dimension specifically, it may have a lower combined score (novelty + viability + fit) than other ideas that excel in viability and fit. The top-ranked shortlist is ordered by total score, meaning the non-obvious pick could rank anywhere within that list or even outside the top K results if its viability or fit scores are low.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}