Bernt
6989a98d75
- Arkitektur: docs/auth/passwordless-architecture.md - Backend: iom/quixzoom-auth-service/ (FastAPI + Redis) - Webb: quixzoom-market-pages/se/login/ (QR-kod + polling) - App: iom/quixzoom-app/src/features/auth/ (push + deep links) Flöde: QR-kod → app-godkännande → webb-inloggad
86 lines
5.9 KiB
HTML
86 lines
5.9 KiB
HTML
<!DOCTYPE html>
|
|
<html lang="en">
|
|
<head>
|
|
<meta charset="UTF-8">
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|
<title>How the Consensus Engine Works | Landvex</title>
|
|
<meta name="description" content="The Landvex Consensus Engine does not vote. It weighs. Learn how multiple observation streams are combined into a single, confidence-scored finding.">
|
|
<link rel="canonical" href="https://landvex.com/insights/how-the-consensus-engine-works/">
|
|
<meta property="og:title" content="How the Consensus Engine Works | Landvex">
|
|
<meta property="og:description" content="The Landvex Consensus Engine does not vote. It weighs. Learn how multiple observation streams are combined into a single, confidence-scored finding.">
|
|
<meta property="og:type" content="article">
|
|
<meta property="og:url" content="https://landvex.com/insights/how-the-consensus-engine-works/">
|
|
<script type="application/ld+json">
|
|
{
|
|
"@context": "https://schema.org",
|
|
"@type": "Article",
|
|
"headline": "How the Landvex Consensus Engine Works: From Observations to Findings",
|
|
"description": "The Landvex Consensus Engine does not vote. It weighs. Learn how multiple observation streams are combined into a single, confidence-scored finding.",
|
|
"author": {"@type": "Organization", "name": "Landvex"},
|
|
"publisher": {"@type": "Organization", "name": "Landvex", "logo": {"@type": "ImageObject", "url": "https://landvex.com/apple-touch-icon.png"}},
|
|
"datePublished": "2026-07-20",
|
|
"dateModified": "2026-07-20",
|
|
"mainEntityOfPage": {"@type": "WebPage", "@id": "https://landvex.com/insights/how-the-consensus-engine-works/"}
|
|
}
|
|
</script>
|
|
<link rel="stylesheet" href="/assets/css/landvex-complete.css">
|
|
<style>
|
|
body { font-family: -apple-system, BlinkMacSystemFont, 'Inter', sans-serif; line-height: 1.6; max-width: 760px; margin: 0 auto; padding: 24px; color: #1a1a1a; }
|
|
h1 { font-size: 2.5rem; font-weight: 700; margin-bottom: 0.5em; line-height: 1.2; }
|
|
h2 { font-size: 1.5rem; font-weight: 600; margin-top: 2em; margin-bottom: 0.5em; }
|
|
p { margin-bottom: 1.2em; }
|
|
.meta { color: #666; font-size: 0.9rem; margin-bottom: 2em; }
|
|
.cta { background: #f5f5f5; padding: 24px; border-radius: var(--radius-md); margin: 2em 0; }
|
|
.cta a { color: #0066cc; text-decoration: none; font-weight: 600; }
|
|
footer { margin-top: 4em; padding-top: 2em; border-top: 1px solid #eee; font-size: 0.85rem; color: #666; }
|
|
</style>
|
|
</head>
|
|
<body>
|
|
<article>
|
|
<h1>How the Landvex Consensus Engine Works: From Observations to Findings</h1>
|
|
<p class="meta">Published July 20, 2026 · Data Quality</p>
|
|
|
|
<p>When multiple contributors observe the same asset, the naive approach is to take a vote. Three people say the lamp is working, one says it is not — majority wins. The problem is that votes treat every input as equally reliable, which is never true.</p>
|
|
|
|
<p>The Landvex <strong>Consensus Engine</strong> takes a different approach. It does not vote. It weighs.</p>
|
|
|
|
<h2>Five signals, one finding</h2>
|
|
<p>Every observation that enters the engine carries metadata beyond the raw answer. The Consensus Engine combines five weighted signals into a single scored finding:</p>
|
|
|
|
<p><strong>1. AI model confidence.</strong> The underlying detection model's certainty about what it sees — a crack, a working lamp, an obscured sign.</p>
|
|
|
|
<p><strong>2. Human validation.</strong> Where available, expert review of a sample of observations, used to calibrate model performance per category and condition.</p>
|
|
|
|
<p><strong>3. Historical consistency.</strong> Whether the current observation aligns with the known history of the same asset. A lamp reported as broken for the third time in two months carries different weight than a first-time report.</p>
|
|
|
|
<p><strong>4. Cross-contributor agreement.</strong> How closely independent observations of the same asset agree, weighted by each contributor's established reliability profile.</p>
|
|
|
|
<p><strong>5. Official record alignment.</strong> Where official GIS or maintenance registers exist, the engine factors in whether the observation confirms or contradicts the registered state.</p>
|
|
|
|
<h2>Weighted aggregation, not averaging</h2>
|
|
<p>The engine does not simply average the five signals. Each signal's weight is adjusted dynamically based on the category, the asset type, and the available data quality. For a lighting asset, model confidence and cross-contributor agreement may dominate. For a vegetation assessment, historical consistency and official record alignment may matter more.</p>
|
|
|
|
<p>The output is not a binary yes/no. It is a <strong>confidence-scored finding</strong>: <em>"Asset condition: degraded, confidence 87%, based on 4 observations, 2 validators, 1 contradiction with register."</em></p>
|
|
|
|
<h2>What happens when signals conflict</h2>
|
|
<p>Conflict is information, not noise. When the AI predicts "intact" and a trusted contributor reports "damaged," the engine flags the asset for priority re-observation rather than forcing a resolution. The contradiction itself becomes a finding: <em>"Condition uncertain — conflicting evidence, requires verification."</em></p>
|
|
|
|
<p>This design avoids the most dangerous failure mode of consensus systems: false certainty. An unresolved conflict is more useful than a wrongly resolved one.</p>
|
|
|
|
<h2>Traceability by design</h2>
|
|
<p>Every finding produced by the Consensus Engine retains a full provenance chain: which observations contributed, what each signal scored, how the weights were applied, and what the confidence calculation yielded. A customer who questions a finding can trace it back to source.</p>
|
|
|
|
<div class="cta">
|
|
<strong>See how consensus-scored findings apply to your decision.</strong><br>
|
|
<a href="/enterprise/">Request a pilot →</a>
|
|
</div>
|
|
|
|
</article>
|
|
|
|
<footer class="lv-footer" role="contentinfo">
|
|
<p><strong>Related:</strong> <a href="/insights/crowdsourced-data-quality/">Can You Trust Crowdsourced Data?</a> · <a href="/insights/decision-first-intelligence/">Decision-First Intelligence</a> · <a href="/methodology/">Our Methodology</a></p>
|
|
</footer>
|
|
|
|
</body>
|
|
</html>
|