Open Questions And Known Gaps

Highest Priority

1. Is server-level cron installed and running?
   The code creates persistent jobs for scoring and OSINT rechecks, but the host scheduler must be confirmed. Without cron, due jobs may only run opportunistically.

2. What is the final Stripe/payment flow?
   Config fields and eligibility logic exist, but full checkout, saved payment method UI, subscription access, webhooks, refunds, and payout policy are not complete.

3. Should low-quality users be blocked, allowed unscored, or heavily suppressed?
   Current code can block posting when payment is required, but product policy around unscored/suppressed posting remains unresolved.

4. What should happen when a writer drops below a paywall/essay threshold?
   Current privileges are calculated from current writer quality, but subscription/premium consequences need a policy.

5. What is the moderation workflow?
   Spam/toxicity scoring exists, but a full moderation queue/dashboard/appeal flow is not built.

Scoring Questions

- Should UI color bands be aligned exactly with verdict thresholds?
- Should confidence penalties be more visible in the UI?
- Should source quality be stronger in the regular score when source URLs are provided?
- Should scores decay or weight recent posts more heavily for writer quality?
- Should writer quality use all posts, only top-level posts, or separate averages by type?
- Should image-only posts have a different score display label?
- Should API cost metadata remain visible behind $, move to admin-only, or be removed?

OSINT Questions

- Is the current truth-factor formula final?
- Should verification speed be rewarded, or can that incentivize rushing?
- Should dispute signals cap at 35, or should AI/source reviews determine the cap?
- How should manual-source abuse be throttled?
- Should OSINT publisher track records be public, partly public, or admin-only?
- Should OSINT track record require 50 resolved posts before displaying a score?
- Should older research be blocked only by text heuristics, or should there be an explicit composer mode?

Trust Score Gap

notes/trust score 1.1.txt is implemented as the active regular AI scoring directive. The older notes/trust score 1.md remains a legacy reference and is not the active spec.

The older full trust-score system is still not implemented.

Missing pieces include:

- user_trust_scores
- trust_events
- device/IP risk tables
- action-specific risk engine
- decaying penalties
- policy mapping from trust bands to platform friction
- admin/moderation dashboard for trust events

Adjacent systems exist:

- content quality scoring
- spam/toxicity flags
- email verification fields
- writer limits
- AI cost caps

But those are not the full trust score.

Paywall/Subscription Gaps

- Full Stripe checkout is not built.
- Subscription access enforcement is not complete.
- Subscriber-only rendering is not fully implemented.
- Public/private subscription activity rules are undecided.
- Platform percentage and payouts are undecided.
- Refund and chargeback rules are undecided.

Testing Gaps

- Remote signed-in browser pass is still important for posting, reposting, OSINT source submission, admin rescore, and paywall UI.
- Database-dependent behavior needs remote temporary PHP tests because no local DB is available.
- SMTP/live email flows need confirmation.
- OSINT cron should be tested after server scheduler setup.

Documentation Gaps

- The technical wiki and FAQ should be kept in sync with these memory docs.
- The memory docs are currently scoring-focused. Other areas may need their own durable memory folders later, such as auth, deployment, payments, and OSINT operations.