Growth Laws (Moore, Koomey, Bell, Zimmerman)
Context: FIT1043_MOC · the exponential trends behind Big Data’s growth · four named laws · why “big” is a moving target
Quick Revision
- 🎯 Objective: explain the exponential growth of computing/data ➔ four laws (Moore, Koomey, Bell, Zimmerman).
- 📦 Core Components: Moore (transistors) | Koomey (energy efficiency) | Bell (computer classes) | Zimmerman (surveillance).
- ⚡ Key Constraint: Koomey and Bell are corollaries of Moore’s Law — efficiency and new device classes follow from doubling transistor counts.
📝 How It Works
1. Moore’s Law (Gordon Moore, Intel, 1965)
- Claim ➔ transistors per chip double every ~2 years (from 1975) ➔ more/bigger/faster memory and CPUs (smaller = faster).
- Nature ➔ microprocessor growth is exponential; the pace is currently slowing.
2. Koomey’s Law (Jonathan Koomey, Stanford, 2010)
- Claim ➔ computations per joule of energy double every ~1.57 years — energy efficiency doubles.
- Consequence ➔ battery needed falls ~100× per decade ➔ ubiquitous computing (a corollary of Moore’s Law).
3. Bell’s Law (Gordon Bell, DEC, 1972)
- Claim ➔ roughly every decade a new, lower-priced computer class forms on a new platform/network/interface ➔ new usage + a new industry.
- Confirmed ➔ PCs, mobile computing, cloud, IoT.
4. Zimmerman’s Law (Phil Zimmermann, 2013, PGP creator)
- Claim ➔ surveillance constantly increasing, privacy constantly decreasing.
⚖️ Core Decision Matrix
| Law | Who / year | What grows | Rate / effect |
|---|---|---|---|
| Moore | Moore, 1965 | transistors per chip | ×2 every ~2 years |
| Koomey | Koomey, 2010 | computations per joule | ×2 every ~1.57 years |
| Bell | Bell, 1972 | new computer classes | one per ~decade |
| Zimmerman | Zimmermann, 2013 | surveillance (privacy ↓) | continuous |
When It Flips: Moore drives the others — doubling transistors (Moore) implies more compute per joule (Koomey) and cheaper new device classes (Bell); those together fuel the data explosion that makes "big" a moving target.
🧠 Active Recall
Contrast Moore's and Koomey's laws, and state what each doubles and how often.
Answer
- Short answer: Moore’s law: transistors per chip double every ~2 years (compute power); Koomey’s law: computations per joule double every ~1.57 years (energy efficiency).
- Why: Power vs efficiency ➔ Moore is about raw transistor count; Koomey (a corollary) is about doing more computation per unit of energy, enabling mobile/ubiquitous devices.
What does Bell's Law predict, and give confirming examples.
Answer
- Short answer: Roughly every decade a new, lower-priced class of computer emerges on a new platform/network/interface, creating new uses and industries — e.g. PCs, mobile, cloud, IoT.
- Why: Cheaper compute (Moore/Koomey) ➔ each cost/efficiency step unlocks a new device category and market.