Your Feed Journey Simulator
Experience how your feed will drift over time, even if you don't change anything.
What Is The Gravitational Pull Effect?
Your interests won't stay balanced. Even if you start with 60% AI and 40% Cooking, you'll drift to 76% AI and 24% Cooking over 6 months—without consciously changing your behavior. This isn't a bug, it's how multiplicative scoring works.
The Core Mechanism
The algorithm uses multiplication, not addition, to score tweets. Your dominant interest gets a scoring advantage, which means you see more of it, which means you engage more with it, which increases the advantage, creating a feedback loop.
Multiplicative scoring (what actually happens):
AI tweet: base_score × 0.60 = higher score
Cooking tweet: base_score × 0.40 = lower score
Result: You see more AI → engage more with AI → AI interest increases → cycle repeatsWhy This Matters
- Inevitable drift: You can't maintain balance without active intervention
- Echo chambers by design: The algorithm concentrates your interests over time
- Loss of diversity: Minority interests get progressively buried
- Unconscious shift: Most users never realize this is happening to them
The Shape of the Drift
The drift is exponential at first, then plateaus. The first 12 weeks see rapid change (60/40 → 70/30), then it slows as you approach saturation (~80/20 is typical plateau). The algorithm doesn't show you what you want—it shows you what it predicts you'll engage with.
Simulate Your Own Journey
Your Starting Point
When you first joined X, you followed a mix of accounts. Let's say you followed accounts in two interest areas. What was your initial split?
The Technical Details
How This Simulator Works
This simulator models the gravitational pull effect based on the actual algorithm code:
- Multiplicative scoring: Cluster interest multiplies tweet scores at candidate generation AND ML scoring stages
- Engagement reinforcement: More visibility → more engagement → higher interest score
- L2 normalization: Your interests sum to 1.0, so one interest increasing means others decrease (zero-sum game)
- FRS acceleration: Follow recommendations create a triple reinforcement (see content → engage → follow)
- Weekly batches: InterestedIn scores update weekly via batch jobs, not real-time
Simplifications in This Model
This simulator uses simplified formulas for illustration. The actual algorithm:
- Assigns you to ~10-20 clusters (out of 145,000 total), not just 2
- Uses complex engagement prediction models with 6,000+ features per tweet
- Applies multiple filters and penalties that compound
- Updates in weekly batches, creating 0-7 day lag
Technical Details
Multiplicative scoring: ApproximateCosineSimilarity.scala:94
InterestedIn calculation: InterestedInFromKnownFor.scala:26-30
Weekly batch updates: InterestedInFromKnownFor.scala:59 - val batchIncrement: Duration = Days(7)
L2 normalization: SimClustersEmbedding.scala:59-72