Skip to main content
Deep Questions with Cal Newport

Beware of “Efficiency Theater”

72 min episode · 3 min read

Episode

72 min

Read time

3 min

Topics

Career Growth, Productivity, Investing

AI-Generated Summary

Key Takeaways

  • ✓Bounded vs. Arbitrary Efficiency: John Smeaton's 1759 water wheel experiments established efficiency as a precise ratio — useful work divided by maximum potential work — capped at 100%. When 20th-century managers applied this concept to workers, they lost the mathematical ceiling entirely. Theorist Walter Polakov named this distinction in his 1921 book: bounded efficiency has physical limits; arbitrary efficiency ratios like tasks-per-employee have none, creating endless pressure to do more.
  • ✓The Knowledge Work Efficiency Trap: The dominant implicit metric in knowledge work — visible tasks completed divided by number of employees — is an arbitrary efficiency ratio with two critical flaws. First, visible task completion correlates loosely, sometimes negatively, with actual value creation. Second, having no upper bound drives workers toward perpetual busyness. Newport labels this dynamic "efficiency theater": performing visible work rather than producing outcomes that generate real organizational value.
  • ✓AI Accelerates the Wrong Metric: AI tools function as arbitrary efficiency accelerators, rapidly inflating task-completion ratios without improving actual output. This explains why Deloitte's research found mixed ROI despite speed gains, and why Goldman Sachs reduced employee access to advanced coding tools like Claude Code despite capability improvements. Deploying AI to maximize task throughput without redefining what constitutes valuable output compounds the original measurement error rather than solving it.
  • ✓Effectiveness as the Alternative Framework: Effectiveness, the measurement framework that efficiency displaced during the industrial revolution, focuses solely on valuable output produced regardless of method. Newport recommends managers conduct monthly narrative reviews where employees make a case for what they produced and why it mattered — no task counts, no email metrics. This shifts evaluation from activity visibility to demonstrated value, making the connection between work and organizational outcomes explicit and testable.
  • ✓The AI Boomerang Phenomenon: Companies cutting staff citing AI productivity gains are rehiring laid-off workers after discovering AI-inflated task ratios don't replicate specific human expertise. The AP documented this pattern across multiple firms. Newport argues this occurs because arbitrary efficiency metrics treat all tasks as interchangeable, while actual organizational capability depends on specific, differentiated expertise that task-count ratios cannot capture or replace through volume increases alone.

What It Covers

Cal Newport traces the mathematical history of efficiency from 18th-century engineer John Smeaton's water wheel experiments through thermodynamics to modern knowledge work, arguing that AI tools are accelerating a fundamental measurement error — optimizing for visible task completion rather than actual value production — and proposes replacing efficiency metrics with effectiveness-based evaluation.

Key Questions Answered

  • •Bounded vs. Arbitrary Efficiency: John Smeaton's 1759 water wheel experiments established efficiency as a precise ratio — useful work divided by maximum potential work — capped at 100%. When 20th-century managers applied this concept to workers, they lost the mathematical ceiling entirely. Theorist Walter Polakov named this distinction in his 1921 book: bounded efficiency has physical limits; arbitrary efficiency ratios like tasks-per-employee have none, creating endless pressure to do more.
  • •The Knowledge Work Efficiency Trap: The dominant implicit metric in knowledge work — visible tasks completed divided by number of employees — is an arbitrary efficiency ratio with two critical flaws. First, visible task completion correlates loosely, sometimes negatively, with actual value creation. Second, having no upper bound drives workers toward perpetual busyness. Newport labels this dynamic "efficiency theater": performing visible work rather than producing outcomes that generate real organizational value.
  • •AI Accelerates the Wrong Metric: AI tools function as arbitrary efficiency accelerators, rapidly inflating task-completion ratios without improving actual output. This explains why Deloitte's research found mixed ROI despite speed gains, and why Goldman Sachs reduced employee access to advanced coding tools like Claude Code despite capability improvements. Deploying AI to maximize task throughput without redefining what constitutes valuable output compounds the original measurement error rather than solving it.
  • •Effectiveness as the Alternative Framework: Effectiveness, the measurement framework that efficiency displaced during the industrial revolution, focuses solely on valuable output produced regardless of method. Newport recommends managers conduct monthly narrative reviews where employees make a case for what they produced and why it mattered — no task counts, no email metrics. This shifts evaluation from activity visibility to demonstrated value, making the connection between work and organizational outcomes explicit and testable.
  • •The AI Boomerang Phenomenon: Companies cutting staff citing AI productivity gains are rehiring laid-off workers after discovering AI-inflated task ratios don't replicate specific human expertise. The AP documented this pattern across multiple firms. Newport argues this occurs because arbitrary efficiency metrics treat all tasks as interchangeable, while actual organizational capability depends on specific, differentiated expertise that task-count ratios cannot capture or replace through volume increases alone.
  • •Jaron Lanier's Effectiveness Model: Computer scientist Jaron Lanier's approach to AI tooling illustrates the effectiveness alternative. Rather than using general AI to maximize output volume, Lanier built a custom tool specifically designed to improve the quality of ideas in his material science work, then evaluated it by whether it produced better solutions. This outcome-first design — build for a specific valuable result, measure that result directly — contrasts with deploying AI broadly to increase visible activity.

Notable Moment

Newport argues that knowledge workers may be inadvertently protected from AI-driven job displacement by their own industry's poor measurement practices. Because firms optimize for the wrong metrics, AI gets deployed ineffectively, limiting its ability to genuinely replace human roles. Better effectiveness measurement could paradoxically accelerate workforce disruption.

Know someone who'd find this useful?

Episode Transcript

So we're at a confusing moment of the AI revolution right now. On the one hand, it seems like every week, the major AI companies have some sort of super impressive new announcement to make of a of a new capability that their tools are able to master. And yet at the same time, we're getting like the recent report from Deloitte, which said when they studied companies deploying AI tools, well, they found that most companies had their workers speed up by using AI tools. The return on investment was mixed, which might explain, for example, why Goldman Sachs recently actually reduced the number of their employees who were given access to advanced coding tools like Claude Code, citing both cost and productivity hits. So what's going on here? How can we have at the same time tools that are demonstrably becoming more impressive in their capabilities and yet we're not seeing this translate in the workplace as cleanly as we would hope? It's not just simply making us all much better at our jobs. Well, today, I want to suggest both an explanation for what's going on and a suggestion for how to fix the problems. And in making these arguments, I'm going to draw from an unlikely source, the history of an incredibly fundamental notion in the modern world, that of efficiency. This is something that we've all heard before, but very few of us know the actual scientific and mathematical history. But if we learn it, we can actually gain a lot of insight about what's going on right now with AI in the workplace and how to make things better. It's a... Trust me, it works. So stick with me here. Let's get into it. Alright. So I'm gonna have to get in the professor mode here because the history of efficiency is something that's going to unlock a lot of insight for us into what is actually happening with AI and the workplace. Let me tell you the source that got me started down this path. It was a 2009 paper. I have it on the screen here for people who are watching. It was written by the historian of science, Jennifer K Alexander, and the title of the paper was The Concept of Efficiency, A Historical analysis. All right. So we can take that back down, Jesse. But that is the paper. I just want to give credit where credit is due that brought me into the world of the history of the notion of efficiency and all the ideas we're going to talk about started from this paper. All right. So what is efficiency? Well, this feels like common sense to all of us. Yeah. You want to be efficient versus not efficient, but it's actually a very complicated concept that has evolved over time. As Alexander points out, it goes all the way back to Aristotle. He didn't use that term, but the notions of efficiency back in Aristotle that then …

Get the full transcript (13,250 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all Deep Questions with Cal Newport transcripts →

You just read a 3-minute summary of a 69-minute episode.

Get Deep Questions with Cal Newport summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

Books, tools, and gear mentioned in this episode

SignalCast may earn commission on purchases via these links. As an Amazon Associate, SignalCast earns from qualifying purchases.

Books

  • by Walter Polakov

    “Theorist Walter Polakov named this distinction in his 1921 book: bounded efficiency has physical limits; arbitrary efficiency ratios like tasks-per-employee have none, creating endless pressure to do more.”

Tools

  • by Anthropic

    “Goldman Sachs reduced employee access to advanced coding tools like Claude Code despite capability improvements.”
  • by Gusto

    “Sponsor: Gusto listed at https://gusto.com/deep”
  • by Pipedrive

    “Sponsor: Pipedrive listed at https://pipedrive.com/deep”
  • by My Body Tutor

    “Sponsor: My Body Tutor listed at https://mybodytutor.com”

Gear

More from Deep Questions with Cal Newport

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

Explore Related Topics

This podcast is featured in Best Mindset Podcasts (2026) — ranked and reviewed with AI summaries.

Read this week's Investing & Markets Podcast Insights — cross-podcast analysis updated weekly.

You're clearly into Deep Questions with Cal Newport.

Every Monday, we deliver AI summaries of the latest episodes from Deep Questions with Cal Newport and 192+ other podcasts. Free for one show.

Start My Monday Digest

No credit card · Unsubscribe anytime