This episode explores two distinct, yet interconnected, constraints that define modern productivity and cognitive health: technology integration and human working memory.

We begin by examining the AI productivity paradox highlighted by a METR study and analyzed by Cal Newport. While AI tools were predicted to deliver massive speedups, experienced developers were found to be 20% slower when engaged in "cybernetic collaboration"—a constant back-and-forth of prompting and reviewing that fragments attention. We propose an alternative approach called "Cybernetic Amplification," where AI is used to eliminate mechanical tasks and cognitive interruptions (like looking up syntax or documentation), thereby preserving the focus intensity necessary for deep work. This approach enabled one non-professional developer to rapidly launch 8 iOS apps in a single summer by accelerating the "Speed-to-Insight Loop".

Next, we zoom out to the ultimate constraint: the human brain's working memory limit of 7±2 items, as defined by George Miller. We discuss how exceeding this capacity leads to cognitive overload in modern life, where we juggle dozens of projects, roles, and notifications. The key to managing complexity isn't maximizing output, but learning to "chunk" related items into manageable units, allowing us to design a more focused and sustainable 7±2 life. The conversation highlights how accepting constraints, whether in design philosophy or cognitive capacity, ultimately leads to clarity and liberation.

Timestamps and links:

00:01:20 - George Miller's "magical number seven" is the basis for DigTek's app MemoryAnchor - https://apps.apple.com/no/app/memory-anchor/id6751170216

00:14:05 - This is a reference to DigTek, the Norwegian indie developer that launched 11 iOS apps during the summer of '25 - https://apps.apple.com/us/developer/andre-berg-st%25C3%25B8len/id1829516490

Disclaimer:

Episodes are based on human-written scripts from essays, design docs, and research. Scripts are AI-refined, creator-approved, then voiced using Google NotebookLM. This is human-directed, AI-assisted storytelling—not AI-generated content. Every idea originates from the creator's work and vision.

Error and Disclaimer 2:

- Timestamp 1:20: METR is wrongfully attributed as Microsoft Engineering Tools Research Study. METR is short for Model Evaluation & Threat Research, metr.org

- When we in the podcast discuss George Miller’s 1956 research on the limits of human memory, proposing that the average person can hold about 7 (plus or minus 2) items in their short-term or working memory, also referenced as "the magical number seven," we are making simplifications in order to get a point across. We encourage anyone interested to do their own deep dive into the actual research and newer studies that add nuance to the picture.

Credits where credits are due:

- Timestamp 2:50: «Cybernectic collaboration» is a term coined by Cal Newport

Relevant articles, info and resources:

- On our blog at digtek.app we have written about our take on Cyberntic Amplification https://digtek.app/blog-2025-09-15-cybernetic-amplification.html

- Also on our blog a post on the 7±2 rule, https://digtek.app/blog-2025-06-11-memoryanchor.html. A follow-up post are in the plans

- Cal Newport on the Deep Life Podcast episode 370 on Cal’s «Cybernetic Collaboration»-term https://www.thedeeplife.com/podcasts/episodes/ep-370-deep-work-in-the-age-of-ai/

You’ll find other blog posts discussing similar topics under «Blog» at https://digtek.app. Our book, «Life as User Experience» is loosely referenced throughout the episode.

Catch you in the next episode!