Building Ungameable Trust Networks for Curated Directories: Decentralized Filtering Without Central Authority
Year: 2026
Type: article
Abstract: Search and discovery on the 2026 internet is broken for three groups: users who want to find real businesses and resources, small businesses who want to be found, and honest affiliate operators who want to earn from genuine recommendations. Search engines rank results by signals that reward scale, not quality. A local hardware store cannot outrank a content farm that produces ten thousand SEO-optimized pages per day with large language models. Paying for search advertising puts small businesses into bidding wars against companies with orders-of-magnitude larger budgets. Social media demands constant content production — videos, posts, engagement farming — work that has nothing to do with the actual service the business provides. Review platforms are corrupted. Fake reviews, review bombing, pay-to-play visibility schemes, and extortion dynamics have made star ratings unreliable. Users append "reddit" to their search queries as a trust signal, but Reddit itself is degrading as it monetizes and as marketers infiltrate communities. AI answer engines attempt to solve the problem by synthesizing responses, but they draw from the same polluted index and introduce hallucination risk. The underlying data is compromised. No amount of summarization fixes a contaminated source. Honest affiliate operators face the same structural disadvantage. An operator who genuinely tests products and recommends what they believe in cannot compete on volume against content farms that rewrite product descriptions and stuff affiliate links without touching the products. The algorithm cannot distinguish the two. The content farm wins on output volume every time. Three intermediaries sit between a person and a useful resource: algorithmic ranking that rewards manipulation, centralized moderation that makes uniform visibility decisions for all users, and review systems that are themselves a manipulation surface. All three are failing.
Author: Howland, Geoffrey
Language: English
Topic: Spam and Phishing Detection
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225065017