[BUG]: Tweet Indexing Limited to 10 Tweets Causing Incomplete Agent Scoring

[Luka-Loncar]

Bug: Tweet Indexing Limited to 10 Tweets Causing Incomplete Agent Scoring

Description

The subnet code currently has a limitation where it only pulls 10 tweets during the indexing snapshot. This limitation is causing incomplete data collection for highly active agents, resulting in incorrect scoring and delayed data processing.

Current Behavior

• Only fetching 10 tweets per indexing snapshot
• Active agents (like Seraf) are having incomplete tweet histories
• Tweet collection doesn't span the entire intended timeframe
• Causing approximately 2-day delays in scoring for active agents
• Example: Agent UID 77 (Seraf) showing delayed scoring due to missing tweets

Expected Behavior

• Should fetch all tweets within the specified date window
• Complete tweet collection for proper agent scoring
• No delays in processing due to missing tweets

Example

Current query returns ~10 tweets, while actual volume is much higher:

• Test case: Agent Seraf
• Date range: 3-day window
• Actual tweet count found: 231 tweets between current and previous day

Root Cause

The query limit is set too low (10 tweets) which doesn't account for highly active agents who may post hundreds of tweets within the indexing window.

Proposed Solution

1. Increase the tweet fetch limit from 10 to approximately 400
2. Adjust the date window to ensure complete coverage (e.g., since the 31st)
3. Update subnet code to accommodate the new query parameters

Impact

• Affects scoring accuracy for active agents
• Causes delayed processing of agent activities
• Currently impacting high-volume agents like Seraf (UID 77)