I've been thinking, following the Dunbar logic, and probably the expansion of the subject of our tribal relationship ranges. I have something for you, I say its a good starting point.
Here’s a simple, transparent “Curation‑Quality Score” on a 1–10 scale. It rewards:
- Breadth – how many different authors you vote for
- Outreach – how much voting weight you send outside your core circle
- Anti‑self‑vote – penalizes heavy self‑voting
1. Definitions (over last 7 days)
- Let Wₜₒₜ = total vote‑weight you’ve cast (sum of all your vote percentages).
- Let U = number of unique authors you’ve voted for.
- Sort those authors by vote‑weight received, pick the top 50.
- Let W₅₀ = sum of vote‑weight you sent to those top 50 authors.
- Let Wₛ = vote‑weight you cast on yourself.
2. Sub‑scores
- Breadth (B) = min(U ÷ 50, 1.0)
- 0 if you only voted 0 authors; 1.0 if you voted ≥ 50 distinct authors.
- Outreach (O) = (Wₜₒₜ − W₅₀) ÷ Wₜₒₜ
- fraction of your weight going outside your top 50.
- Anti‑Self (S) = 1 − (Wₛ ÷ Wₜₒₜ)
- 1.0 if you never self‑vote; decreases toward 0 as self‑voting grows.
Each sub‑score runs from 0 to 1.
3. Combine & Scale
Take a weighted average (equal weights here):
RawScore = (B + O + S) / 3
Convert to 1–10:
CurationScore = round( RawScore * 9 ) + 1
So RawScore=0→score 1; RawScore=1→score 10.
4. Examples
| Curator | U (# authors) | W₅₀ / Wₜₒₜ | Wₛ / Wₜₒₜ | B = min(U/50,1) | O | S | RawScore | 1–10 Score |
|---|---|---|---|---|---|---|---|---|
| A (narrow) | 10 | 100% | 20% | 0.2 | 0.00 | 0.80 | (0.2+0+0.8)/3 = 0.33 | 0.33×9+1 ≈ 3 → 3 |
| B (core‑only) | 50 | 100% | 0.00 | 1.0 | 0.00 | 1.00 | (1+0+1)/3 = 0.67 | 0.67×9+1 ≈ 7 → 7 |
| C (diverse) | 100 | 25% | 0.00 | 1.0 | 0.75 | 1.00 | (1+0.75+1)/3 = 0.92 | 0.92×9+1 ≈ 10 → 10 |
| D (self‑heavy) | 30 | 60% | 0.40 | 0.60 | 0.40 | 0.50 | (0.6+0.4+0.5)/3 = 0.50 | 0.50×9+1 = 5.5 → 6 |
- A votes only 10 authors, all weight to those → low breadth & outreach
- B hits 50 authors but no outreach beyond them → middling
- C votes 100 authors, most weight outside top 50 → near perfect
- D mixes some outreach but heavy self‑vote drags score down
5. How to implement
- Pull your last‑7‑day vote history via Hive API.
- Compute Wₜₒₜ, tally weight per author, count U, identify top 50 and W₅₀, measure self‑vote Wₛ.
- Calculate B, O, S, then RawScore and map to 1–10.
- Display “CurationScore: X/10” alongside a mini‑chart of your B, O, S sub‑scores.
That gives you a quick, intuitive metric of how thoughtfully you’re curating beyond your inner circle—rewarding exploration, penalizing collusion, and still allowing for healthy core‑community support.
RE: I unfollowed an account I used to upvote recently