Description
I have been using the plugin the entire time while playing normally to contribute data. Initially, I was detected as ~90% real player.
After continuing normal gameplay (training + questing), this dropped to ~69% → ~64%, which seems incorrect.
My experience
I was doing normal activities:
- Grinding MM2 caves (range XP)
- Spider cave using Lunar spell "Cure Me" (high click spam) Magic XP
- Completing Dragon Slayer II for the first time
During DS2:
Vorkath (quest only):
- Day 1: ~6 deaths
- Day 2: break
- Day 3: after 11 deaths → first kill
Next day:
- Marked as ~4% "Vorkath bot"
Important:
- I only have 1 Vorkath kill (quest kill)
- No farming at all
Final boss:
- Multiple deaths while learning
- Killed on 4th attempt
Next day:
- Bot probability increased by ~9%
Problem
Normal player behavior is being flagged as bot-like:
- Learning mechanics (repeated deaths)
- Re-gearing and retrying
- High click-rate manual training
- Quest-based bossing (low kill count)
Even with full gameplay history (including mistakes), the model trends toward bot classification.
Expected
- Questing should be treated differently from farming
- Low kill-count bosses should not increase bot probability
- Deaths and inconsistent play should reduce bot likelihood
Summary
Legitimate gameplay (questing, learning bosses, training) is being misclassified as bot-like. The model seems overly sensitive to repetition and input patterns without enough context.
Description
I have been using the plugin the entire time while playing normally to contribute data. Initially, I was detected as ~90% real player.
After continuing normal gameplay (training + questing), this dropped to ~69% → ~64%, which seems incorrect.
My experience
I was doing normal activities:
During DS2:
Vorkath (quest only):
Next day:
Important:
Final boss:
Next day:
Problem
Normal player behavior is being flagged as bot-like:
Even with full gameplay history (including mistakes), the model trends toward bot classification.
Expected
Summary
Legitimate gameplay (questing, learning bosses, training) is being misclassified as bot-like. The model seems overly sensitive to repetition and input patterns without enough context.