Five Core Metrics

LAWRENCE H. PUTNAM, SR., a leading expert in the software estimation and management fields, is president emeritus and founder of Quantitative Software Management, a software management consulting firm based in McLean, Virginia. Larry and coauthor Ware Myers wrote three other books and numerous articles together over many years. WARE MYERS was an independent consultant and a long-time contributing editor to Computer and IEEE Software magazines. Prior to his death in 2011, Ware's professional interests included the application of metrics to software planning, estimating, bidding, and project control.

&>



Introduction 3



The Evolution of the Metrics 5

Computing Nuclear Weapons Effects 6

Computer Budgets in the Pentagon 7

Applying the Rayleigh Concept to New Projects 10

The Rayleigh Concept Leads to the Software Equation 12

Second Key Relationship: Manpower Buildup Equation 15

The Rayleigh Curve As a Process Control Vehicle 16




Part I What Software Stakeholders Want 19




Chapter 1 Some Software Organizations Are Doing Very Well 21

Maturity Assessments Reflect Hope 24

Has Software Development Mired Down? 26

Way Down Low 27

Intelligence Can Guide Us 28




Chapter 2 A Finite Planet Makes Measurement Essential 31

What Makes Metrics Effective? 33

What the Right Metrics Are 33

How the Core Metrics Relate 34

The Management of Software Projects Is Very Difficult 34

Specifications 35

Capability Maturity Model 35

Process 36

Measure What Has Been Done 36

These Five Metrics Work Together 37




Chapter 3 Integrate Metrics with Software Development 40

Metrics Meter Limited Resources 40

What Is the Process? 42

Phase One: Inception 44

Phase Two: Elaboration 45

Phase Three: Construction 48

Phase Four: Transition 50




Chapter 4 "I Want Predictability" 52

The Software Situation Is Serious 53

Department of Defense Calls for Predictability 54

The Underlying Reasons Are Complex 54

Competition Stirs the Pot 55

The Limits of the Possible 56

There Are Certain Limits 60




Part II The Metrics Needed for Effective Control 63




Chapter 5 The Measurement View 65

The Profound Difference 67

One Fact Is Minimum Development Time 68

The Core Measurements Underlie "Getting Work Done" 68

The Key Concepts 69

Expressing the Key Concepts in Metrics 71

Time 71

Effort 71

Quality 72

Amount of Work 72

Process Productivity 73




Chapter 6 Estimating Size as a Measure of Functionality

Shannon's Path 78

Representing the Amount of Functionality 81

Sizing Functionality by Calibration 82

Sizing Implications of Reuse 83

Estimate Size 84

Get the Facts 85

Get the People 86

Get the Data 86

Allow Time 86

Employ an Estimating Method 86




Chapter 7 Penetrating the Software Productivity Jungle 89

Finding the Right Relationship 90

The Relationship of Time to Size 90

The Relationship of Effort to Size 91

The Software Equation 92

The Effect on Productivity 92

Conventional Productivity Varies Widely 93

Schedule Is a Factor in Productivity 94

How Conventional Productivity Behaves 94

How Process Productivity Behaves 97

Obtain Process Productivity by Calibration 98

Don't Do It This Way! 101

Do It with Process Productivity 101






Chapter 8 Defect Rate Measures Reliability 103



The Fifth Core Metric 104

Software Development Is in Trouble 106

Why Do We Commit Errors? 107

Plan Projects to Minimize Error-Making 107

Limit Functionality 108

Allow Sufficient Schedule Time 108

Allow Sufficient Effort 109

Improve Process Productivity 109

Find Defects 110

Some Defects Remain at Delivery 110




Part III Control at the Project Level 113




Chapter 9 Do the Hard Stuff First-Establish Feasibility 115

The Vision Comes First 116

What to Build 116

Risk 117

Economic Constraints 117

An Endless Task 117

How We Got into This Leaky Boat 119

The Inception Phase Establishes Feasibility 120

Delimit Scope 121

Select an Architecture Possibility 122

Mitigate Critical Risk 123

A Feasibility Decision, 150 Years Later 123

Make the Business Case 124

Where Do the Resources Come From? 126

Were Our Ancestors Shrewder? 127




Chapter 10 Do the Tough Stuff Next-Functional Design 128

The Tough Stuff 129

Meshing the Activities 131

Formulating Requirements Is No Longer Simple 132

Difficulties Fixing Requirements 133

What We Need to Do 134

Key Requirements Lead to Functional Design 134

Identify Significant Risks? 135

The Business Case at Phase Two 136

Supporting Phase two Itself 136

Nearing the Bid Decision 137




Chapter 11 The Power of the Trade-Off 138

Avoid the Impossible Region 140

Stay Out of the Impractical Region 141

Trade-Off Time and Effort 141

Trade-Off Time and Defects 142

Small Is Beautiful 143

Once More with Emphasis: Small Is Beautiful! 144

Selecting the Software Relationship 151

There Must Be a Relationship 152

The Relationship Must Be Reasonably Accurate 153

Management Must Use the Relationship 156

The Bid Is Different from the Estimate 158






Chapter 12 Turning Your Range Estimate Into Your Client's Point Bid 159



The Uncertainty Drivers 160

The Time-Effort Curve Is Uncertain 160

Estimators Face Constraints 161

Manage Estimating Risk 163

Get Out of the Uncertainty Trap 164

Work with Uncertainty, Not in Defiance of It 166




Chapter 13 The Main Build Leads to Operational Capability 169

Staff Allocation 171

Projecting Functionality 175

Detecting Defects in Test 177

Apply Statistical Control to the Core Metrics 177

Track System Size 180

Track Defects 182

The Essential Point: Control 185




Chapter 14 The Operation and Maintenance Phase 186

Where Does Phase Four Begin? 187

What Goes On in the Phase? 188

Find and Fix Defects 189

Improve Reliability 189

Adapt the System 190

Modify the System 190

When Does the Phase End? 191

How to Estimate the Phase 192




Chapter 15 Replan Projects in Trouble 195

The Award Differs from the Original Estimate 196

The Award Reflects the Estimate 199

Replan Troubled Projects 200

Replan Periodically 201






Part IV Control at the Organization Level 203






Chapter 16 A Telecommunications Company Evaluates Its Software Suppliers 205

Lessons from Construction 206

Lack of Facts Leads to Pain 206

Lower Tier, the Same 208

PTT Telecom BV Gets Real 209

The PTT Faces Competition 210

The Telecom Considers What to Do 210

How to Do It? 211

Purchasing Alerts Management 212

The Telecom Gets Results 213

Results on Sample Projects 215




Chapter 17 Evaluate Bids on the Facts 217

The Reality Is Research and Development 217

Getting to the Facts 218

The Bidder Cooperates 219

The Evaluators Consider Schedule 219

The Evaluators Consider Effort 222

The Evaluators Consider Reliability 223

The World of Wild Bidding 223

The Vendor Tries to Buy In 224

The Vendor Gold Plates 225

Experience Supports the Factual Approach 225

Who Needs the Facts? 226




Chapter 18 Managing a Portfolio of Projects 227

A Senior Manager Can Master-Plan 228

Good Project Plans Underlie a Good Master Plan 229

What a Master Plan Does 232

The Master Curve Shows Declining Staff 232

The Master Curve Shows Increasing Staff 233

Control of Project Backlog 233

Coping with Powerful Project Managers 234

Planning Staff at the Enterprise Level 234




Chapter 19 Improving the Software Development Process 236

Performance Comes from the Entire Process 237

Pre-Phase: Needs 238

Inception Phase (or Feasibility) 239

Elaboration Phase (or Functional Design) 240

Construction Phase (or Main Build) 241

Transition (or Operation and Maintenance) 241

Phases Are Implemented by Workflows 241

Process Improvement Comes Hard 242

Process Productivity Rises Slowly 243

Process Productivity Extends Over a Great Range 243

Why Has Process Improvement Been Hard? 244

Trapped in the Economy 245

Need for Client Participation 245

Software Development Is Difficult 246

Some Organizations Are Improving Their Process 247

The Place to Start Is Where You Are 250




Chapter 20 Managing Reusable Components 251

The Five Stages of Reuse 252

Hip-Pocket Reuse 252

Reuse from a Repository 252

Product-Line Reuse 254

ERP Systems Are a Form of Reuse 255

Architecture-Wide Reuse 255

What Are the Essential Supports for Reuse? 257

Architecture Accommodates Components 257

Interfaces Enable Inter-Component Communication 258

Variation Mechanisms Multiply Adaptability 258

Process Provides Locations for Selection 258

Modeling Language Provides the Means 259

Tools Implement These Needs 260

Internet-Hosted Services 260

Estimating in the Age of Reuse 260

Components from a Repository 261 Product-Line and ERP Reuse 261

Any Metric Measuring Functionality Will Do 261

Finding That Metric 261

Calibration to the Rescue 262

Fundamental Principle Reinforced 263

Architecture-Wide Components Increase Reuse 264

Functionality or Process Productivity? 264

Size Adjustment Seeks "Effective" Size 266

Second Approach Adjusts Process Productivity 266




Chapter 21 Metrics Backstop Negotiation 269

Negotiation Bridges Gaps 270

The Principal Gaps 272

Different Interests at Stake 273

A Job for Negotiation Man! 273

The Core Metrics Support Negotiation 274

Software Development Depends on Negotiation 276

Negotiation Depends Upon Metrics 277

Chapter 22 Motivating Participants in the Software Development Process 278

People Are the Ultimate Resource 279

Hire Good People 279

Build Organizations 280

Keep People 280

Give People Time 281

Five Essential Motivating Factors 282

Extending Estimating Wisdom More Widely 282

Ten Great Truths 284

The Five Great Truths About Metrics 284

The Five Great Truths About Software Development 285




Appendix A Good Enough Is Better Than Floundering 287






Appendix B Behavior of the Core Metrics 290








Study Results 291

Size 292

Effort 293

Time 294

Staff 295

Process Productivity 296

Mean Time To Defect 297

Will the Next Period Reverse Again? 298

Bibliography 301

Index 305