Add author reputation signal to trending algorithm

[realproject7]

Summary

Fixes #288

• Add computeAuthorReputation() combining five sub-signals: follower_count (Farcaster, log-scaled), x_followers_count (Twitter, log-scaled), x_verified (boolean boost), neynar_score (0-1), and quotient_score (0-1000 normalized)
• Integrate as 6th trend signal at 20% weight; rebalance existing signals (rating 20%, price 15%, TVL 15%, continuation 15%, recency 15%)
• Batch-fetch writer user rows via primary_address + verified_addresses (overlaps query) to avoid N+1 — two parallel queries regardless of storyline count
• Gracefully returns 0 reputation for writers with no user record

Test plan

• Verify trending page loads without errors
• Confirm storylines by writers with higher reputation scores rank higher (all else equal)
• Check that storylines by unknown writers (no user row) still appear with 0 reputation penalty
• Validate no N+1 queries in server logs (only 2 user queries total)

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
plotlink	Ignored		Mar 27, 2026 6:07am

[project7-interns]

Verdict: APPROVE

Summary

The PR matches issue #581: it adds a composite author reputation signal, integrates it as the sixth trending input at 20% weight, and keeps the user lookup batched instead of introducing N+1 queries.

Findings

• No blocking issues found.

Decision

Approve. The implementation is scoped correctly, follows the existing Supabase/address-normalization patterns, and the ranking changes are coherent with the acceptance criteria.