novari
  • Explore
  • Live
  • Social
  • Theory
TrendingSportsPoliticsCorporate & ESGMacroCulture
NovariNovari

The global consensus measurement platform. Expertise-weighted forecasts and collective intelligence, distilled into the Novari Score.

Product

  • Home
  • Explore
  • Live

Community

  • Social
  • Profile
  • Leaderboard

Company

  • About
  • Theory
  • Terms
  • Privacy

© 2026 Novari. All rights reserved.

TermsPrivacy
ExploreLiveSocialTheory

novari Theory

How millions of independent voices become one reliable number

The Novari Score distills what the world believes into a single, transparent number. Here's the thinking behind it — no statistics degree required.

The Power of Collective Intelligence

The Consummation of Empire by Thomas Cole (1836) — a civilization at its peak, built not by any single mind but by the accumulated wisdom of generations. New-York Historical Society, public domain.

In 1907, the statistician Francis Galton attended a county fair where visitors guessed the weight of an ox. Nearly 800 people entered — farmers, butchers, and people who had never seen an ox up close. When Galton averaged all the guesses, the crowd's answer was almost perfect: off by less than one percent. Most individuals were wildly wrong. But together, they were better than any single expert.

This isn't a fluke. Decades of research have shown that large groups consistently outperform individuals — but only when four conditions are met.

The most valuable thing a consensus platform produces is not money — it is a number that captures what the world actually believes.

Four Conditions for Crowd Wisdom

  • Diversity of opinion — People bring different knowledge, experiences, and perspectives. A group of clones won't outperform a single clone.
  • Independence — Each person forms their own view without copying others. When people anchor to what everyone else said, the group gets dumber.
  • Decentralization — No single gatekeeper controls the answer. Local expertise from around the world contributes to the final score.
  • Good aggregation — There's a mechanism to combine individual voices into a group answer. A crowd isn't wise if nobody is counting.

Novari is built on these principles. The next section explains how we turn millions of individual voices into a single, reliable number.

How the Novari Score Works

The Tholos at Delphi, Greece — the ancient Oracle of Delphi represents humanity's oldest tradition of seeking truth through structured inquiry. Wikimedia Commons, public domain.

Think of it like a teacher grading a class. Students who consistently get the right answers earn more influence when the class needs to reach a group answer. That's the core idea behind the Novari Score: your track record determines your voice.

Three Key Ideas

  1. Everyone participates — Anyone can weigh in on any question. There are no gatekeepers. Every voice starts equal.
  2. Accuracy builds influence — Over time, the system tracks how often each person's predictions match reality. People who are consistently right earn more weight in the consensus.
  3. The score emerges from weighted voices — The Novari Score isn't set by editors or algorithms. It emerges naturally from the combined input of all participants, weighted by their demonstrated accuracy.

From Votes to Score

Here is the pipeline that turns individual votes into the published Novari Score. Each step adds a layer of quality.

Raw VotesIndividual rankings submittedExpertise LookupE(u) from historical accuracyWeight Assignmentw(u) proportional to expertiseWeighted AggregationS(a) = sum of w(u) × R(a)Quality CheckSE and m_eff thresholdsNCSNovari Consensus Score
End-to-end pipeline from raw votes to the published Novari Consensus Score.

How Topics Mature

Every new topic starts with editorial input — a curated seed that gives the question structure. As more people participate, the community gradually takes over. Once enough votes come in, the score is fully community-driven. The transition is smooth — there's never a sudden quality drop.

Editorial SeedingCurated initial rankingsParticipation growsBlended TransitionCommunity gradually takes overThreshold reachedFull CommunityExpertise-weighted consensusbeta = 0beta = 1beta = min(1, N/T) controls the editorial-to-community blendOrange marker indicates the transition threshold
Three-phase lifecycle from editorial seeding to full community-driven consensus.

Built to Be Trustworthy

The Reading Room of the British Museum — a monument to accumulated knowledge and the tradition of open scholarly inquiry. Wikimedia Commons, public domain.

A score is only useful if you can trust it. Here are the four biggest threats to any consensus system — and how the Novari Score handles each one.

What Could Go Wrong?

“What if people guess randomly?”

People who guess randomly will be wrong often. Their accuracy score drops, and their influence on the consensus approaches zero. The system naturally filters out noise by listening more to people who have demonstrated they know what they're talking about.

“What if someone creates fake accounts?”

New accounts start with near-zero influence. You can't game the system by creating a thousand accounts — each one would need to build a genuine track record of accurate predictions before it carries any weight. Historical accuracy is the proof of legitimacy.

“What if people copy each other?”

For opinion-based questions, results are hidden until you vote. You can't be influenced by a score you haven't seen. This ensures each vote reflects an independent belief, not a bandwagon effect.

“What if someone tries to game it?”

Imagine a top forecaster deliberately submitting a wrong answer to move the score. Their accuracy drops. Their influence drops. Future attempts to manipulate become even less effective. Gaming the system is self-defeating — the more you try, the less power you have.

The System Improves Itself

The Novari Score gets better over time through a feedback loop: accurate participants gain influence, which improves the consensus, which raises the bar for accuracy, which further rewards the best forecasters. No manual curation needed — the system self-corrects.

User submits rankingCycle closes, truth revealedACCURATE PREDICTIONINACCURATE PREDICTIONHigh accuracy scoreExpertise E(u) risesHigher weight next cycleLow accuracy scoreExpertise E(u) fallsWeight asymptotes to zeroStrategic or dishonest voters cannot sustain influence:accuracy drops → expertise falls → weight drops in future cycles.
Self-correcting feedback loop: accuracy determines expertise, which determines consensus influence.

The Oracle Network

Jantar Mantar astronomical observatory, Jaipur — monumental instruments for measurement and prediction, embodying the long arc of scientific inquiry. Wikimedia Commons, public domain.

Imagine 10,000 people weigh in on a question. The community consensus lands at 67%. But the top 5% of proven forecasters — people with hundreds of accurate predictions — say 81%. That 14-point gap is the story. It's the tension between what everyone thinks and what the most accurate people think.

Two Scores, One Question

The Oracle Network produces a separate consensus from top-tier forecasters. When it disagrees with the broader community, you get the most interesting content on the platform: “The experts say X, but the crowd says Y.” That tension drives discussion, sharing, and deeper engagement.

Oracle Network ScoreTop 5% expertise-weightedCommunity ConsensusAll voters equally weightedDeliberate divergencePublished NCS
Dual-score architecture: Oracle Network vs. Community Consensus converge into the published Novari Consensus Score.

Five Levels of Expertise

Both volume and accuracy are required to advance. You can't climb the ranks by making hundreds of random guesses — you have to be consistently right.

  1. Novice — Fewer than 10 predictions. Gray badge. Everyone starts here.
  2. Analyst — 25+ predictions, top 50% accuracy. Blue badge. Visible in the Experts tab.
  3. Strategist — 100+ predictions, top 20% accuracy. Silver badge. Featured in category feeds.
  4. Oracle — 250+ predictions, top 5% accuracy. Gold badge. Votes carry extra weight in the consensus.
  5. Superforecaster — 500+ predictions, top 2% accuracy. Platinum badge. Named in the Oracle Network.
Novice< 10 predictionsAnalyst25+ predictions, top 50%Strategist100+ predictions, top 20%Oracle250+ predictions, top 5%Superforecaster500+ predictions, top 2%
Five-tier expertise hierarchy based on prediction volume and accuracy percentile.

What Comes Next

Charles Babbage's Difference Engine — the mechanical ancestor of modern computation, bridging historical calculation with today's machine learning. Wikimedia Commons, public domain.

The Novari Score turns millions of independent voices into a single, transparent number. It rewards accuracy over volume, self-corrects over time, and resists manipulation by design. But this is just the foundation.

As data accumulates and the community grows, the system gets smarter. Here's what we're building next.

Looking Ahead

  • Smarter expertise tracking — Instead of fixed tiers, the system will learn a continuous expertise function from prediction history. Your influence adapts in real time as your accuracy changes.
  • Consensus trajectory forecasting — Predicting where the score is heading before voting closes, using patterns from previous cycles. Think of it as a weather forecast for collective opinion.
  • Personalized recommendations — Surfacing questions you're most likely to have an informed opinion on, based on your prediction style and accuracy patterns.

The Core Idea

When millions of diverse, independent people share what they believe — and the system listens more to those who are consistently right — you get something remarkable: a continuously updated, self-correcting measure of collective intelligence. That's the Novari Score.

1997Surowiecki FoundationWisdom of Crowdsframework forcollectiveintelligence2000Prediction MarketsEarly marketexperimentsdemonstrate forecastaccuracy2005Tetlock's Expert JudgmentSuperforecastingresearch validatescalibrated expertise2008Continuous ForecastingAlways-on forecastingsystems gain traction2026NovariSocial-first consensuswithexpertise-weightedrankings
Key milestones in collective intelligence research leading to the Novari platform.