Steam, Steel, and Infinite Minds: Are We Repeating the Mistakes of Factory Owners 100 Years Ago?
Executive Summary
Yes—we are repeating the mistakes of factory owners from 100 years ago. Drawing on Notion founder Ivan Zhao's essay *Steam, Steel and Infinite Minds* (source), this post argues that most AI applications today are "bolted onto" existing workflows, delivering only modest gains rather than a genuine productivity explosion. The real breakthrough requires breaking free of the "water source constraint"—the limits of human cognitive and communication bandwidth—and moving from tool substitution to system redesign.
The Historical Metaphor: The "Waterwheel Phase"
Lessons from the Industrial Revolution
Early textile mills in Lancashire, England were built by rivers and powered by waterwheels. When steam engines arrived, most factory owners took the conservative "swap out the waterwheel" approach: they replaced the wheel in place, keeping the same buildings, layouts, processes, and riverside locations. The result: only modest gains, far below steam's transformative potential.
The breakthrough came from forward-thinking owners who decoupled from water entirely—relocating factories near workers, ports, and raw materials, and redesigning production around the steam engine's properties.
| Waterwheel-Swap Phase | System-Redesign Phase | |---|---| | Kept riverside sites | Free of geographic constraints | | Kept original layouts | Redesigned production flows | | Simple power substitution | Distributed power configuration | | Modest gains | Productivity explosion |
Today's AI Paradigm Error
> "We're still in the 'swap out the waterwheel' phase."
The typical pattern: AI chatbots bolted onto existing software—AI windows in CRMs, writing assistants in editors, copilots in collaboration platforms. Meanwhile, modern corporate structures trace back to 19th-century railway companies' need to coordinate thousands of people, all built on the implicit assumption that humans are the only cognitive agents.
Three-Level Transformation Framework
Individual: From "Bicycle" to "Self-Driving"
Building on Steve Jobs' "bicycles for the mind" metaphor, Zhao describes an evolution path:
| Stage | Power Source | Multiplier | Mode | |---|---|---|---| | Bicycle | Human effort | 1-10x | Individual executes all tasks | | 10x engineer | High skill | 10x | Deep expertise | | 30-40x engineer | AI orchestration | 30-40x | Multi-agent management, async queues | | Self-driving | Autonomous systems | 100x+ | AI-led, human-supervised |
Case in point: Notion co-founder Simon, already a known "10x programmer," now rarely writes code himself—instead orchestrating three or four AI agents simultaneously, reaching a 30-40x engineer level of productivity.
Organizational: AI as "Steel"
Current corporate communication infrastructure is like building skyscrapers out of wood—relying on human brains connected through meetings and messages, which collapses as organizations scale. Every decision layer needs alignment meetings; every cross-team effort needs multi-level approvals.
> "AI is steel for organizations."
Like steel, AI can maintain contextual continuity across workflows and surface decisions precisely without adding noise:
- Noise reduction: 2-hour meetings → 5-minute async reviews
- Decision speed: three-level approvals → minutes
- Scale without decay: growth without efficiency loss
Economic: From Florence to Tokyo
Centuries ago, cities were human-scaled—you could walk across Florence in forty minutes; pace of life was set by walking distance and voice range. Today's knowledge economy is a "Florence": teams capped at dozens, workflows paced by meetings and email, organizations breaking down beyond a few hundred people.
Steel and steam changed cities themselves: steel frames enabled skyscrapers, railways connected centers to hinterlands, and megacities like Tokyo, Chongqing, and Dallas emerged—not as "bigger Florences" but as fundamentally different ways of living: disorienting and anonymous, yet offering more opportunity, freedom, and combinatorial possibility.
| Current: "Sparse" organizations | Future: "Dense" organizations | |---|---| | Human time lost to meetings, coordination, waiting | Agents running 24/7 | | Context switching drains cognition | Humans focus on key decisions and creative breakthroughs | | Efficiency decays with scale | 1,000 humans + 10,000 agents | | Bound by biology and geography | Cross-timezone, multilingual, global operations |
This redefines the boundary of a "company": a 100-person company could run 10,000 agents, delivering the service capacity of a traditional tens-of-thousands-person firm.
Breaking Through Barriers
The core obstacles are not technical but cognitive and organizational: escaping "waterwheel thinking" means redesigning workflows around AI's strengths rather than grafting AI onto human-designed processes.
Conclusion
The argument echoes the post's opening thesis: as with steam and steel, the productivity explosion will not come from swapping components but from those willing to rebuild systems from scratch—moving from tools that assist humans to organizations where humans and "infinite minds" (AI agents) work as one continuous, densely connected whole.
*Source: Ivan Zhao, "Steam, Steel and Infinite Minds" — Notion blog*