Fractal Geometry of Software Design

Presented at Craft 2025

Transcript

1. coupling.dev #Cra ft Conf vladikk THE FRACTAL GEOMETRY OF SOFTWARE DESIGN
2. VLAD KHONONOV @vladikk vladikk.com coupling.dev
3. coupling.dev #Cra ft Conf vladikk
4. coupling.dev #Cra ft Conf vladikk • How much food do we need to stay alive? • How creative are we? • How profitable are our companies?
5. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. How much food do we need to stay alive? How creative are we? How profitable are our companies?
6. coupling.dev #Cra ft Conf vladikk METABOLIC RATE OF ANIMALS West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
7. coupling.dev #Cra ft Conf vladikk REGISTERED PATENTS West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
8. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. INCOME AND ASSETS OF COMPANIES
9. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. How much food do we need to stay alive? How creative are we? How profitable are our companies?
10. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
11. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. How much food do we need to stay alive? How creative are we? How profitable are our companies?
12. coupling.dev #Cra ft Conf vladikk "Viewed through this lens, cities, companies, plants, animals, our bodies, and even tumors manifest a remarkable similarity in the ways that they are organized and function." - Geo ff rey West
13. coupling.dev #Cra ft Conf vladikk AGENDA 1. A primer on energy supply networks 2. How network theory applies to software design 3. Leveraging the network theory to manage complexity 4. S ^ D
14. coupling.dev #Cra ft Conf vladikk ENERGY SUPPLY NETWORKS
15. coupling.dev #Cra ft Conf vladikk ENERGY SUPPLY NETWORKS Systems that are sustained by the transport of energy through hierarchical networks.
16. coupling.dev #Cra ft Conf vladikk ENERGY SUPPLY NETWORKS
17. coupling.dev #Cra ft Conf vladikk ENERGY SUPPLY NETWORKS
18. coupling.dev #Cra ft Conf vladikk SOFTWARE DESIGN AS AN ENERGY SUPPLY NETWORK
19. coupling.dev #Cra ft Conf vladikk WHAT IS THE ENERGY SUPPLIED BY SOFTWARE DESIGN?
20. coupling.dev #Cra ft Conf vladikk WHAT IS THE ENERGY SUPPLIED BY SOFTWARE DESIGN? — KNOWLEDGE
21. coupling.dev #Cra ft Conf vladikk
22. coupling.dev #Cra ft Conf vladikk Shared Knowledge Encapsulated Knowledge Boundaries Encapsulated Knowledge
23. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling
24. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling • How the parts of the system are implemented • Private interfaces • Not intended for inter-component integration
25. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling • Knowledge of what the system has to do • Business logic • Use cases • Rules, invariants, constraints
26. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling • Models of the business domain • Data structures • Software system-related aspects of the business domain • Understanding of the functional requirements
27. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling • Components’ integration contracts • Integration-specific models, APIs • The minimal data needed for components to work together
28. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN So ft ware systems are sustained by the transport of knowledge (energy) through hierarchical networks of modules.
29. coupling.dev #Cra ft Conf vladikk ECONOMIES OF SCALE IN ENERGY SUPPLY NETWORKS
30. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
31. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
32. coupling.dev #Cra ft Conf vladikk METABOLIC RATE OF ANIMALS West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
33. coupling.dev #Cra ft Conf vladikk METABOLIC RATE OF ANIMALS West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
34. coupling.dev #Cra ft Conf vladikk y = 3.4 * x3/4 West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
35. coupling.dev #Cra ft Conf vladikk y = 3.4 * x3/4 y = x Sublinear Growth Linear Growth Sublinear function grows slower than any linear function
36. coupling.dev #Cra ft Conf vladikk Chihuahua Yorkshire Terrier Pug Beagle Boxer Newfoundland 1.8kg 3.5kg 7kg 14.5kg 29kg 58kg
37. coupling.dev #Cra ft Conf vladikk Chihuahua Yorkshire Terrier Pug Beagle Boxer Newfoundland 1.8kg 3.5kg 7kg 14.5kg 29kg 58kg BASAL METABOLIC RATE The minimum number of calories a day needed to stay alive without performing any activities. https://www.sciencedirect.com/topics/engineering/basal-metabolic-rate
38. coupling.dev #Cra ft Conf vladikk Chihuahua Yorkshire Terrier Pug Beagle Boxer Newfoundland 1.8kg 3.5kg 7kg 14.5kg 29kg 58kg 110 220 440 880 1,760 3,520 Breed Weight Basal Metabolic Rate
39. coupling.dev #Cra ft Conf vladikk Chihuahua Yorkshire Terrier Pug Beagle Boxer Newfoundland 1.8kg 3.5kg 7kg 14.5kg 29kg 58kg 110 220 440 880 1,760 3,520 110 180 302 521 876 1,474 Breed Weight Linear Thinking Basal Metabolic Rate
40. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. NO. OF GASOLINE STATIONS
41. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
42. coupling.dev #Cra ft Conf vladikk REGISTERED PATENTS West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
43. coupling.dev #Cra ft Conf vladikk y = x1.15 y = x West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson.
44. coupling.dev #Cra ft Conf vladikk y = x1.15 y = x Superlinear Growth Linear Growth Superlinear function grows faster than any linear function
45. coupling.dev #Cra ft Conf vladikk ECONOMIES OF SCALE Superlinear Growth Sublinear Growth Linear Growth
46. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN So ft ware systems are sustained by the transport of knowledge (energy) through hierarchical networks of modules.
47. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Functionality (Linear growth)
48. coupling.dev #Cra ft Conf vladikk ECONOMIES OF SCALE Superlinear Growth Sublinear Growth Linear Growth
49. coupling.dev #Cra ft Conf vladikk ECONOMIES OF SCALE Superlinear Growth Sublinear Growth Linear Growth WHY DON’T ENERGY SUPPLY NETWORKS GROW INDEFINITELY?
50. coupling.dev #Cra ft Conf vladikk GROWTH LIMITS
51. coupling.dev #Cra ft Conf vladikk "You can plainly see the impossibility of increasing the size of structures to vast dimensions either in art or in nature; … also it would be impossible to build up the bony structures of men, horses, or other animals so as to hold together and perform their normal functions . . . for if his height be increased inordinately he will fall and be crushed under his own weight.” - Galileo Galilei
52. coupling.dev #Cra ft Conf vladikk
53. coupling.dev #Cra ft Conf vladikk
54. coupling.dev #Cra ft Conf vladikk Cross-section Resistance to fracture is proportional to the cross-sectional area
55. coupling.dev #Cra ft Conf vladikk x2 x2 x2 Weight increase: Strength increase: 2 * 2 * 2 = x8 2 * 2 = x4
56. coupling.dev #Cra ft Conf vladikk
57. coupling.dev #Cra ft Conf vladikk x2 x2 x2 Weight increase: Strength increase: 2 * 2 * 2 = x8 2 * 2 = x4
58. coupling.dev #Cra ft Conf vladikk GROWTH LIMITS Superlinear Growth Sublinear Growth Linear Growth • As a system grows, it becomes more efficient • Undesired consequences of growth become more efficient as well
59. coupling.dev #Cra ft Conf vladikk GROWTH LIMITS IN SOFTWARE SYSTEMS
60. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Functionality (Linear growth)
61. coupling.dev #Cra ft Conf vladikk M8 M10 M9 M6 M1 M2 M3 M4 M5 M7
62. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Functionality (Linear growth)
63. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth)
64. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth) "You can plainly see the impossibility of increasing the size of structures to vast dimensions either in art or in nature; … so also it would be impossible to build up the bony structures of men, horses, or other animals so as to hold together and perform their normal functions . . . for if his height be increased inordinately he will fall and be crushed under his own weight.” - Galileo Galilei
65. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth)
66. coupling.dev #Cra ft Conf vladikk EXTENDING GROWTH LIMITS
67. coupling.dev #Cra ft Conf vladikk “I have sketched a bone whose natural length has been increased three times and whose thickness has been multiplied until, for a correspondingly large animal, it would perform the same function which the small bone performs for its small animal” - Galileo Galilei
68. coupling.dev #Cra ft Conf vladikk “if one wishes to maintain in a great giant the same proportion of limb as that found in an ordinary man he must find a harder and stronger material for making the bones” - Galileo Galilei
69. coupling.dev #Cra ft Conf vladikk POSTPONING GROWTH LIMITS 1. Adapt form / shape to withstand increased stress 2. Use stronger materials
70. coupling.dev #Cra ft Conf vladikk SUSTAINABLE GROWTH REQUIRES INNOVATION
71. coupling.dev #Cra ft Conf vladikk https://en.wikipedia.org/wiki/List_of_largest_cities_throughout_history
72. coupling.dev #Cra ft Conf vladikk https://en.wikipedia.org/wiki/List_of_largest_cities_throughout_history Stone tools, Fire, Bow, Wheel Writing Water Mill Gunpowder Printing Press Optical Lens Steam Engine Telegraph Plastics Airplane Transistor Nuclear Power Space Flight Internet Mobile Technology
73. coupling.dev #Cra ft Conf vladikk INNOVATION IN SOFTWARE DESIGN
74. coupling.dev #Cra ft Conf vladikk “if one wishes to maintain in a great giant the same proportion of limb as that found in an ordinary man he must find a harder and stronger material for making the bones” - Galileo Galilei
75. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth) Cognitive Limit (superlinear growth)
76. coupling.dev #Cra ft Conf vladikk “I have sketched a bone whose natural length has been increased three times and whose thickness has been multiplied until, for a correspondingly large animal, it would perform the same function which the small bone performs for its small animal” - Galileo Galilei
77. coupling.dev #Cra ft Conf vladikk "Highly complex, self-sustaining structures, whether cells, organisms, ecosystems, cities, or corporations, require the close integration of enormous numbers of their constituent units that need e ff icient servicing at all scales. This has been accomplished in living systems by evolving fractal-like, hierarchical branching network systems.” - Geo ff rey West
78. coupling.dev #Cra ft Conf vladikk
79. coupling.dev #Cra ft Conf vladikk
80. coupling.dev #Cra ft Conf vladikk
81. coupling.dev #Cra ft Conf vladikk http://paulbourke.net/fractals/googleearth/
82. coupling.dev #Cra ft Conf vladikk Y. Chen, P. Cheng, Heat transfer and pressure drop in fractal tree-like microchannel nets Lorthois, Sylvie, and Francis Cassot. "Fractal analysis of vascular networks: insights from morphogenesis." Journal of theoretical biology 262.4 (2010): 614-633.
83. coupling.dev #Cra ft Conf vladikk Chihuahua Yorkshire Terrier Pug Beagle Boxer Newfoundland 1.8kg 3.5kg 7kg 14.5kg 29kg 58kg 110 220 440 880 1,760 3,520 110 180 302 521 876 1,474 Breed Weight Linear Thinking Basal Metabolic Rate
84. coupling.dev #Cra ft Conf vladikk West, G. (2018). Scale: The universal laws of life and death in organisms, cities and companies. Weidenfeld & Nicolson. NO. OF GASOLINE STATIONS
85. coupling.dev #Cra ft Conf vladikk Sublinear Growth Linear Growth Sublinear function grows slower than any linear function
86. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth) Complexity Sublinear growth)
87. coupling.dev #Cra ft Conf vladikk FRACTAL MODULARITY
88. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN Intrusive Coupling Functional Coupling Model Coupling Contract Coupling Shared Knowledge Encapsulated Knowledge Boundaries Encapsulated Knowledge
89. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN So ft ware systems are sustained by the transport of knowledge (energy) through hierarchical networks of modules.
90. coupling.dev #Cra ft Conf vladikk INTEGRATION STRENGTH (S) Intrusive Coupling Functional Coupling Model Coupling Contract Coupling Shared Knowledge Encapsulated Knowledge Boundaries Encapsulated Knowledge
91. coupling.dev #Cra ft Conf vladikk DISTANCE (D) Downstream Module Upstream Module depends on Distance
92. coupling.dev #Cra ft Conf vladikk DISTANCE (D) M ethods Classes Sub-M odules M odules Com ponents (M icro)Services System s Coordination
93. coupling.dev #Cra ft Conf vladikk COST OF CHANGE IS PROPORTIONAL TO THE DISTANCE BETWEEN COMPONENTS
94. coupling.dev #Cra ft Conf vladikk Downstream Module Upstream Module depends on Integration Strength (S) Distance (D)
95. coupling.dev #Cra ft Conf vladikk LOW STRENGTH HIGH STRENGTH LOW DISTANCE HIGH DISTANCE
96. coupling.dev #Cra ft Conf vladikk LOW STRENGTH HIGH STRENGTH ? LOW DISTANCE HIGH DISTANCE • High cost of cascading changes • Low shared knowledge • Minimizes cascading changes
97. coupling.dev #Cra ft Conf vladikk LOW STRENGTH HIGH STRENGTH Loose Coupling LOW DISTANCE HIGH DISTANCE • High cost of cascading changes • Low shared knowledge • Minimizes cascading changes
98. coupling.dev #Cra ft Conf vladikk LOW STRENGTH HIGH STRENGTH ? Loose Coupling LOW DISTANCE HIGH DISTANCE • High shared knowledge • Frequent cascading changes • Low cost of cascading changes
99. coupling.dev #Cra ft Conf vladikk LOW STRENGTH HIGH STRENGTH High Cohesion Loose Coupling LOW DISTANCE HIGH DISTANCE • High shared knowledge • Frequent cascading changes • Low cost of cascading changes
100. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH ? Loose Coupling LOW DISTANCE HIGH DISTANCE • Low cost of cascading changes • Low shared knowledge • Co-located unrelated modules
101. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling LOW DISTANCE HIGH DISTANCE • Low cost of cascading changes • Low shared knowledge • Co-located unrelated modules
102. coupling.dev #Cra ft Conf vladikk Local Complexity
103. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling ? LOW DISTANCE HIGH DISTANCE • High cost of cascading changes • High shared knowledge • Frequent cascading changes
104. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE • High cost of cascading changes • High shared knowledge • Frequent cascading changes
105. coupling.dev #Cra ft Conf vladikk Global Complexity
106. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE
107. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE
108. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE MODULARITY = STRENGTH XOR DISTANCE
109. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE M = S XOR D
110. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE M = S ^ D
111. coupling.dev #Cra ft Conf vladikk 1. Minimize integration strength as much as possible 2. Balance integration strength and distance • Both are high? — Reduce distance • Both are low? — Increase distance APPLYING M=S^D
112. coupling.dev #Cra ft Conf vladikk 1. Minimize integration strength as much as possible 2. Balance integration strength and distance • Both are high? — Reduce distance • Both are low? — Increase distance 3. Repeat at all levels of abstraction APPLYING M=S^D
113. coupling.dev #Cra ft Conf vladikk CLASS Method Method Method Method Method
114. coupling.dev #Cra ft Conf vladikk PACKAGE / NAMESPACE Class Class Class Class Class
115. coupling.dev #Cra ft Conf vladikk (MICRO)SERVICE Namespace Namespace Namespace Namespace Namespace
116. coupling.dev #Cra ft Conf vladikk DISTRIBUTED SYSTEM (Micro)service (Micro)service (Micro)service (Micro)service (Micro)service
117. coupling.dev #Cra ft Conf vladikk Intrusive Functional Model Contract INTEGRATION STRENGTH
118. coupling.dev #Cra ft Conf vladikk M = S ^ D
119. coupling.dev #Cra ft Conf vladikk WRAP UP
120. coupling.dev #Cra ft Conf vladikk KNOWLEDGE: THE ENERGY OF SOFTWARE DESIGN So ft ware systems are sustained by the transport of knowledge (energy) through hierarchical networks of modules.
121. coupling.dev #Cra ft Conf vladikk Knowledge (sublinear growth) Complexity (superlinear growth) Cognitive Limit (No growth) Functionality (Linear growth)
122. coupling.dev #Cra ft Conf vladikk High Cohesion LOW STRENGTH HIGH STRENGTH Local Complexity Loose Coupling Global Complexity LOW DISTANCE HIGH DISTANCE M = S XOR D
123. coupling.dev #Cra ft Conf vladikk
124. coupling.dev #Cra ft Conf vladikk
125. COUPLING.DEV
126. coupling.dev #Cra ft Conf vladikk
127. THANK YOU! coupling.dev #Cra ft Conf vladikk