New tech, new smells – Your future job won’t be writing code but understanding and fixing code, often written by AI New tech, new smells – Your future job won’t be writing code but understanding and fixing code, often written by AI TL;DR: You reference external AI conversations to explain code instead of writing declarative tests TL;DR: You reference external AI conversations to explain code instead of writing declarative tests Problems 😔 Comments
External dependencies
Broken links
Unverified behavior
Knowledge fragmentation
Maintenance burden
Lost context
Obsolete Comments
Misleading explanation Comments Comments External dependencies Broken links Broken links Unverified behavior Knowledge fragmentation Maintenance burden Lost context Obsolete Comments Obsolete Comments Misleading explanation Solutions 😃 Write executable tests
Remove external references
Do not blindly trust the AI
Describe with inline examples
Keep tests local
Remove all comments
Replace Magic Numbers with constants. Write executable tests executable tests Remove external references Do not blindly trust the AI Describe with inline examples Keep tests local Remove all comments all comments Replace Magic Numbers with constants. Magic Numbers Refactorings ⚙️ https://maximilianocontieri.com/refactoring-011-replace-comments-with-tests?embedable=true https://maximilianocontieri.com/refactoring-011-replace-comments-with-tests?embedable=true Context 💬 If you add comments that reference external AI conversations, Stack Overflow posts, or online resources to explain how your functions work, you are not thinking about your reader. These references create dangerous external dependencies that break over time. Links become dead, conversations get deleted, and future maintainers cannot access the context they need to understand your code. When you rely on external AI advice instead of writing proper tests, you create code that appears documented but lacks verification and local understanding. The moment you rely on an external AI chat to explain what your code does, you make your codebase dependent on a conversation that might disappear, change, or get outdated. A unit test is more effective than any link. It defines what the code does and what you expect it to do. No need to click or guess. Comments and documentation often lie. Code never does. Sample Code 📖 Wrong ❌ def calculate_starship_trajectory(initial_velocity, fuel_mass, 
                                burn_rate, gravity=9.81):
    """
        
    See explanation at
    https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
         
    """
    # AI suggested this approach
    burn_time = fuel_mass / burn_rate
    
    # Physics formula from Claude conversation
    # https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
    delta_v = gravity * burn_time * 0.85  
    # 0.85 explanation 
    # https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
    final_velocity = initial_velocity + delta_v
    
    # Return format suggested by GPT 
    return {
        'burn_time': burn_time,
        'final_velocity': final_velocity,
        'delta_v': delta_v
    }

def calculate_orbit_insertion(velocity, altitude):
    """
       
    Algorithm explanation available at:
    https://claude.ai/chat/orbit-insertion-help-session
    """
    # See AI conversation for why we use this formula
    orbital_velocity = (velocity * 1.1) + (altitude * 0.002)
    return orbital_velocity def calculate_starship_trajectory(initial_velocity, fuel_mass, 
                                burn_rate, gravity=9.81):
    """
        
    See explanation at
    https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
         
    """
    # AI suggested this approach
    burn_time = fuel_mass / burn_rate
    
    # Physics formula from Claude conversation
    # https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
    delta_v = gravity * burn_time * 0.85  
    # 0.85 explanation 
    # https://claude.ai/share/5769fdd1-46e3-40f4-b9c6-49efbee93b90
    final_velocity = initial_velocity + delta_v
    
    # Return format suggested by GPT 
    return {
        'burn_time': burn_time,
        'final_velocity': final_velocity,
        'delta_v': delta_v
    }

def calculate_orbit_insertion(velocity, altitude):
    """
       
    Algorithm explanation available at:
    https://claude.ai/chat/orbit-insertion-help-session
    """
    # See AI conversation for why we use this formula
    orbital_velocity = (velocity * 1.1) + (altitude * 0.002)
    return orbital_velocity Right 👉 def calculate_starship_trajectory(initial_velocity, fuel_mass, 
                                burn_rate, gravity=9.81):
  
    THRUST_EFFICIENCY = 0.85
    
    burn_time = fuel_mass / burn_rate
    delta_v = gravity * burn_time * THRUST_EFFICIENCY
    # You replace the magic number
    final_velocity = initial_velocity + delta_v
    
    return {
        'burn_time': burn_time,
        'final_velocity': final_velocity,
        'delta_v': delta_v
    }

def calculate_orbit_insertion(velocity, altitude):
    """Calculate orbit insertion velocity."""
    
    VELOCITY_BOOST_FACTOR = 1.1
    ALTITUDE_ADJUSTMENT_RATE = 0.002

    orbital_velocity = (velocity * VELOCITY_BOOST_FACTOR) +
      (altitude * ALTITUDE_ADJUSTMENT_RATE)
    return orbital_velocity    

import unittest
from starship_trajectory_calculator import (
    calculate_starship_trajectory, calculate_orbit_insertion
)

class TestStarshipTrajectoryCalculator(unittest.TestCase):
    
    def test_basic_trajectory_calculation(self):
        result = calculate_starship_trajectory(100, 1000, 10)
        
        self.assertEqual(result['burn_time'], 100.0)
        self.assertEqual(result['delta_v'], 833.85)
        self.assertEqual(result['final_velocity'], 933.85)
    
    def test_zero_fuel_scenario(self):
        result = calculate_starship_trajectory(200, 0, 10)
        
        self.assertEqual(result['burn_time'], 0.0)
        self.assertEqual(result['delta_v'], 0.0)
        self.assertEqual(result['final_velocity'], 200.0)
    
    def test_high_burn_rate(self):
        result = calculate_starship_trajectory(150, 500, 100)
        
        self.assertEqual(result['burn_time'], 5.0)
        self.assertAlmostEqual(result['delta_v'], 41.69, places=2)
        self.assertAlmostEqual(result['final_velocity'], 191.69, 
                             places=2)
    
    def test_custom_gravity(self):
        result = calculate_starship_trajectory(100, 600, 20, 
                                             gravity=3.71)  # Mars
        
        self.assertEqual(result['burn_time'], 30.0)
        self.assertAlmostEqual(result['delta_v'], 94.76, places=2)
        self.assertAlmostEqual(result['final_velocity'], 194.76, 
                             places=2)
    
    def test_orbit_insertion_basic(self):
        orbital_velocity = calculate_orbit_insertion(7800, 400000)
        
        self.assertEqual(orbital_velocity, 9380.0)
    
    def test_orbit_insertion_low_altitude(self):
        orbital_velocity = calculate_orbit_insertion(7500, 200000)
        
        self.assertEqual(orbital_velocity, 8650.0)
    
    def test_orbit_insertion_zero_altitude(self):
        orbital_velocity = calculate_orbit_insertion(8000, 0)
        
        self.assertEqual(orbital_velocity, 8800.0) def calculate_starship_trajectory(initial_velocity, fuel_mass, 
                                burn_rate, gravity=9.81):
  
    THRUST_EFFICIENCY = 0.85
    
    burn_time = fuel_mass / burn_rate
    delta_v = gravity * burn_time * THRUST_EFFICIENCY
    # You replace the magic number
    final_velocity = initial_velocity + delta_v
    
    return {
        'burn_time': burn_time,
        'final_velocity': final_velocity,
        'delta_v': delta_v
    }

def calculate_orbit_insertion(velocity, altitude):
    """Calculate orbit insertion velocity."""
    
    VELOCITY_BOOST_FACTOR = 1.1
    ALTITUDE_ADJUSTMENT_RATE = 0.002

    orbital_velocity = (velocity * VELOCITY_BOOST_FACTOR) +
      (altitude * ALTITUDE_ADJUSTMENT_RATE)
    return orbital_velocity    

import unittest
from starship_trajectory_calculator import (
    calculate_starship_trajectory, calculate_orbit_insertion
)

class TestStarshipTrajectoryCalculator(unittest.TestCase):
    
    def test_basic_trajectory_calculation(self):
        result = calculate_starship_trajectory(100, 1000, 10)
        
        self.assertEqual(result['burn_time'], 100.0)
        self.assertEqual(result['delta_v'], 833.85)
        self.assertEqual(result['final_velocity'], 933.85)
    
    def test_zero_fuel_scenario(self):
        result = calculate_starship_trajectory(200, 0, 10)
        
        self.assertEqual(result['burn_time'], 0.0)
        self.assertEqual(result['delta_v'], 0.0)
        self.assertEqual(result['final_velocity'], 200.0)
    
    def test_high_burn_rate(self):
        result = calculate_starship_trajectory(150, 500, 100)
        
        self.assertEqual(result['burn_time'], 5.0)
        self.assertAlmostEqual(result['delta_v'], 41.69, places=2)
        self.assertAlmostEqual(result['final_velocity'], 191.69, 
                             places=2)
    
    def test_custom_gravity(self):
        result = calculate_starship_trajectory(100, 600, 20, 
                                             gravity=3.71)  # Mars
        
        self.assertEqual(result['burn_time'], 30.0)
        self.assertAlmostEqual(result['delta_v'], 94.76, places=2)
        self.assertAlmostEqual(result['final_velocity'], 194.76, 
                             places=2)
    
    def test_orbit_insertion_basic(self):
        orbital_velocity = calculate_orbit_insertion(7800, 400000)
        
        self.assertEqual(orbital_velocity, 9380.0)
    
    def test_orbit_insertion_low_altitude(self):
        orbital_velocity = calculate_orbit_insertion(7500, 200000)
        
        self.assertEqual(orbital_velocity, 8650.0)
    
    def test_orbit_insertion_zero_altitude(self):
        orbital_velocity = calculate_orbit_insertion(8000, 0)
        
        self.assertEqual(orbital_velocity, 8800.0) Detection 🔍 Automatic Automatic You can detect this smell by searching for comments containing URLs to AI chat platforms, external forums, or references to "AI suggested" or "according to conversation". Look for functions that have detailed external references but lack corresponding unit tests. Exceptions 🛑 Academic or research code might legitimately reference published papers or established algorithms. However, these should point to stable, citable sources and permanent links rather than ephemeral AI conversations, and should still include comprehensive tests. Tags 🏷️ Comments Comments Level 🔋 Beginner Beginner Why the Bijection Is Important 🗺️ In the real world, you don't rely on external authorities to validate your understanding of critical processes. You develop internal knowledge and verification systems. Your code should reflect this reality by containing all necessary understanding within itself through tests and clear implementation. When you break this correspondence by depending on external AI conversations, you create fragile knowledge that disappears when links break or platforms change, leaving future maintainers without the context they need. correspondence Links are not behavior. Tests are. AI Generation 🤖 AI generators sometimes create this smell because they frequently suggest adding references to the conversation or external sources where the solution was previously discussed. They tend to generate excessive comments that point back to their explanations rather than creating self-contained, testable code. AI Detection 🧲 AI can detect this smell when you ask it to identify external references in comments, especially URLs pointing to AI chat platforms. Most AI tools can help convert the external explanations into proper unit tests when given clear instructions. Try Them! 🛠 Remember: AI Assistants make lots of mistakes Remember: AI Assistants make lots of mistakes Suggested Prompt: Replace this external reference and comments with coverage and unit tests Suggested Prompt: Replace this external reference and comments with coverage and unit tests Without Proper Instructions

With Specific Instructions



ChatGPT

ChatGPT



Claude

Claude



Perplexity

Perplexity



Copilot

Copilot



Gemini

Gemini



DeepSeek

DeepSeek



Meta AI

Meta AI



Grok

Grok



Qwen

Qwen Without Proper Instructions

With Specific Instructions



ChatGPT

ChatGPT



Claude

Claude



Perplexity

Perplexity



Copilot

Copilot



Gemini

Gemini



DeepSeek

DeepSeek



Meta AI

Meta AI



Grok

Grok



Qwen

Qwen Without Proper Instructions

With Specific Instructions Without Proper Instructions Without Proper Instructions With Specific Instructions With Specific Instructions ChatGPT

ChatGPT ChatGPT ChatGPT ChatGPT ChatGPT ChatGPT ChatGPT Claude

Claude Claude Claude Claude Claude Claude Claude Perplexity

Perplexity Perplexity Perplexity Perplexity Perplexity Perplexity Perplexity Copilot

Copilot Copilot Copilot Copilot Copilot Copilot Copilot Gemini

Gemini Gemini Gemini Gemini Gemini Gemini Gemini DeepSeek

DeepSeek DeepSeek DeepSeek DeepSeek DeepSeek DeepSeek DeepSeek Meta AI

Meta AI Meta AI Meta AI Meta AI Meta AI Meta AI Meta AI Grok

Grok Grok Grok Grok Grok Grok Grok Qwen

Qwen Qwen Qwen Qwen Qwen Qwen Qwen Conclusion 🏁 External references to AI conversations create fragile documentation that breaks over time and fragments your codebase's knowledge. You should replace these external dependencies with self-contained unit tests that both document and verify behavior locally, ensuring your code remains understandable and maintainable without relying on external resources. Relations 👩‍❤️‍💋‍👨 https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxxvii?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxxvii?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxx https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxx https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxxi https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xxxi https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xv https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-xv https://hackernoon.com/code-smell-259-control-your-environment-to-avoid-test-failures https://hackernoon.com/code-smell-259-control-your-environment-to-avoid-test-failures Disclaimer 📘 Code Smells are my opinion. opinion Credits 🙏 Photo by julien Tromeur on Unsplash julien Tromeur Unsplash The best documentation is code that doesn't need documentation The best documentation is code that doesn't need documentation Steve McConnell Steve McConnell https://hackernoon.com/400-thought-provoking-software-engineering-quotes?embedable=true https://hackernoon.com/400-thought-provoking-software-engineering-quotes?embedable=true This article is part of the CodeSmell Series. https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true https://hackernoon.com/how-to-find-the-stinky-parts-of-your-code-part-i-xqz3evd?embedable=true

The code in this story is for educational purposes. The readers are solely responsible for whatever they build with it.

Keep

Satellites, Sovereignty, and the Next Internet Revolution

Use Infinity Instead of None in Python Loops

Buy My Book

Share your ideas with me!

Read My Stories

Buy Me a Coffee

Book Author

Too Long; Didn't Read

Replace AI References in Comments with Tests That Actually Work

Replace AI References in Comments with Tests That Actually Work

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

10 Predictions About the Future of Finance

The Noonification: Use This 7-Step McKinsey Framework to Solve Any Problem (1/10/2023)

The Noonification: A Taxonomy of Inclusiveness (1/11/2024)

The Noonification: What is the InfiniteNature-Zero AI Model? (11/19/2022)

10 Ways AI Has Changed Our Lives

100 Days of AI, Day 8: Experimenting With Microsoft's Semantic Kernel Using GPT-4

10 Predictions About the Future of Finance

The Noonification: Use This 7-Step McKinsey Framework to Solve Any Problem (1/10/2023)

The Noonification: A Taxonomy of Inclusiveness (1/11/2024)

The Noonification: What is the InfiniteNature-Zero AI Model? (11/19/2022)

10 Ways AI Has Changed Our Lives

100 Days of AI, Day 8: Experimenting With Microsoft's Semantic Kernel Using GPT-4

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps