Desearch releases scoring overhaul with stricter penalties

Desearch deployed a major validator update tightening organic and synthetic scoring through concurrency testing, cheap code checks on all responses, and seven new penalty types. The release splits deep scoring (20% sample) from cheap checks (100%), distributes deep slots by organic volume, and adds strict penalties for response time, summary structure, duplicates, date ranges, and sort order. A separate security concern was raised about sybil protection gaps and IP-spoofing vulnerabilities in the codebase.

• •Concurrency testing now exercises declared miner capacity via batched UID-sorted queries
• •Seven new cheap penalties: summary structure, count, date range, duplicates, schema, sort order, link verification
• •Deep scoring sample reduced to 20% per UID; all organics run cheap checks; slots distributed by volume
• •MinRealisticTime penalty zeros sub-threshold responses outright
• •Security flags: sybil protection bypassed via separate coldkeys; IP-spoofing attack possible in axon declaration