MinMaxHub
Documentation
  • Introduction
  • Architecture
  • Frontend
Domains
  • Nodes
  • Entities
  • Consensus
  • Suggestions
Resources
  • API Reference

Copyright 2026 MinMaxHub

  • Documentation
  • Domains
  • API Reference
NodesEntitiesSuggestionsVotesConsensusUsersRulesMedia
powered by Zudoku
Domains

Consensus

The Consensus domain is the source of truth for MinMaxHub's public content. It bridges the gap between the "Chaos" of community suggestions and the "Order" of public consumption. Consensus stores the finalized, materialized view of every entity page, ready for high-speed retrieval.

How It Works

Consensus follows a two-step process: moderation then computation.

1. Moderation

Admins review suggestions in the Moderation Queue, sorted by confidence score. They can approve or reject individual suggestions. Moderation does not trigger consensus recomputation -- that is a separate, explicit action.

2. Compute Consensus

When an admin triggers consensus computation for an entity, the system:

  1. Collects all eligible suggestions for each section (topic + item) of the entity.
  2. Ranks them using the Wilson Score algorithm, which favors statistically confident quality over raw vote count.
  3. Selects a winner for each section and materializes it as a consensus item.
  4. Removes any previously-existing consensus items that no longer have a winner.
  5. Invalidates the cached static page so the next visitor sees fresh content.

Wilson Score

Rather than ranking by simple vote count (up - down), MinMaxHub uses the Wilson Score Confidence Interval. This answers: "Given these votes, what is the statistical probability that this is a good suggestion?"

This means:

  • A suggestion with 10 up / 0 down ranks higher than 50 up / 40 down (higher confidence).
  • A suggestion with 100 up / 0 down ranks higher than 1 up / 0 down (more data).
  • Ties are broken by oldest submission first.

Serving Public Pages

When a visitor loads a content page:

  1. The system resolves the URL slug to an entity.
  2. All consensus items for that entity are fetched in a single read.
  3. The page is assembled by walking the entity's layout and filling each section with its consensus winner.
  4. Sections with no consensus winner fall back to the layout's default value.

The result includes the entity metadata, layout structure, consensus content, and breadcrumbs -- everything needed to render the page in one response.

Consensus Item

Each consensus item represents the winning suggestion for a specific section of a page:

FieldDescription
entityIdThe entity this consensus belongs to
topicKeyWhich section of the layout
itemKeyWhich item in a collection (null for singletons)
winningSuggestionIdThe suggestion that won
valueThe materialized content
authorId / authorUsernameWho wrote the winning suggestion
confidenceWilson Score
upvotes / downvotesVote tallies from the winning suggestion
computedAtWhen consensus was last computed

API Endpoints

MethodEndpointPurposeAuth
POST/consensus/compute/{entityId}Recompute consensus for an entityRequired
GET/consensus/page/{entityId}Get assembled consensus page by entity IDPublic
GET/consensus/resolve/{slugPath+}Resolve a page by URL slug (includes breadcrumbs)Public
GET/consensus/moderation-queueGet low-confidence items for admin reviewRequired
POST/consensus/moderate/{suggestionId}Approve or reject a suggestionRequired
Last modified on April 2, 2026
VotesUsers
On this page
  • How It Works
    • 1. Moderation
    • 2. Compute Consensus
    • Wilson Score
  • Serving Public Pages
  • Consensus Item
  • API Endpoints