X Algorithm Playbook

Logo

The definitive guide to maximizing your reach on X (Twitter), based on reverse-engineering the open-source algorithm.

View the Project on GitHub tang-vu/x-algorithm-playbook

Avoiding Penalties

How to avoid the negative actions and filters that destroy your reach.


The Negative Action Weights

These actions SUBTRACT from your score:

Action Approximate Weight Impact
Report -20× a like Devastating
Block -10× a like Very severe
Mute -5× a like Severe
“Not Interested” -1× a like Moderate

Reality check: One block undoes the positive signal from 10 likes.


What Triggers Negative Actions

Block Triggers

Behavior Risk Level
Spam replies 🔴 High
Unsolicited DMs 🔴 High
Harassment 🔴 High
Extreme views 🟡 Medium
Over-tagging 🟡 Medium
Off-topic engagement 🟢 Low

Mute Triggers

Behavior Risk Level
Posting too frequently 🔴 High
Off-brand content 🟡 Medium
Engagement bait 🟡 Medium
Repetitive content 🟢 Low

Report Triggers

Behavior Risk Level
ToS violations 🔴 Automatic
Misinformation 🔴 High
Harassment 🔴 High
Spam patterns 🟡 Medium

“Not Interested” Triggers

Behavior Risk Level
Irrelevant content 🟡 Medium
Clickbait 🟡 Medium
Low quality 🟢 Low

The 12 Filters That Can Hide Your Content

Before scoring even happens, these filters can remove your post:

1. Age Filter

2. Drop Duplicates Filter

3. Core Data Hydration Filter

4. Self Tweet Filter

5. Retweet Deduplication Filter

6. Ineligible Subscription Filter

7. Previously Seen Posts Filter

8. Previously Served Posts Filter

9. Muted Keyword Filter ⚠️

10. Author Socialgraph Filter ⚠️

11. VF Filter (Visibility Filtering) ⚠️

12. Dedup Conversation Filter

Most important to avoid: #9, #10, #11


Muted Keywords to Avoid

Common words that get muted:

Promotional Spam

❌ "DM me"
❌ "Link in bio"
❌ "Follow for follow"
❌ "Giveaway" (spam context)
❌ "Free money"

Crypto/Scam Adjacent

❌ "100x"
❌ "Guaranteed returns"
❌ "Not financial advice"
❌ "WAGMI" (depending on audience)

Engagement Bait

❌ "Like if you agree"
❌ "Retweet to win"
❌ "Follow + RT"

Political/Divisive (Niche-Dependent)

⚠️ Political keywords (muted by many)
⚠️ Controversial names
⚠️ Culture war terms

Content Safety Guidelines

Stay Within ToS

✅ Safe ❌ Dangerous
Strong opinions Personal attacks
Debate Harassment
Criticism Threats
Adult topics (marked) Explicit content (unmarked)

Avoid Automated Detection

Patterns that may trigger spam detection:

❌ Same reply on multiple posts
❌ Mass following/unfollowing
❌ Suspicious posting patterns
❌ Coordinated behavior (pods)
❌ Link spamming

Protecting Your Reputation

Engagement Hygiene

✅ Do ❌ Don’t
Engage meaningfully Reply with “Great post!” only
Space your engagement Mass-engage in short bursts
Stay in your niche Randomly engage everywhere
Build relationships Transactional engagement

Handling Conflict

When someone attacks you:
├── Don't engage (engagement signals interest)
├── Block if necessary (protects your experience)
├── Move on (don't create drama)
└── Never escalate (risk getting blocked by others)

Recovery from Penalties

If Your Reach Dropped

  1. Audit recent content - Did you trigger negative actions?
  2. Check for pattern changes - Posting frequency, content type
  3. Wait and observe - Sometimes it’s algorithmic noise
  4. Slowly rebuild - Focus on high-quality, safe content

Signs of Penalty

Symptom Possible Cause
Sudden reach drop Multiple blocks/mutes
Posts not showing Filter triggered
Low engagement on quality content Embedding confusion

Quick Reference

Pre-Post Safety Check

Engagement Safety Check


Next: Growth Strategies →