Pi, deliberate practice, and a small app experiment

During the holidays I became curious about how far deliberate practice can push memory, especially for someone who is objectively bad at memorizing numbers (me).

I had watched Veritasium’s “The Expert Myth,” where Grant Gussman recites 200 digits of π during an interview. The interesting part wasn’t the performance, but his conclusion: “It’s not talent. It’s a system.”

The system aligns with the four well-known conditions for expertise in cognitive science:

Valid environment
Repetition
Fast feedback
Deliberate practice

As someone who builds products, that instantly sounded like a feedback loop design problem rather than a talent problem.

So I decided to test it.

Building the experiment

I built a small iOS app that turns π into a compact practice loop with immediate correctness feedback. Typing digits, getting feedback, repeating the loop, and increasing the index.

It took only a few evenings to prototype.

Results

To my surprise, I reached 160 digits much faster than expected. A few observations from the experiment:

chunking emerges naturally after around 60 digits
the “difficulty” drops as chunk size increases
feedback latency is more important than motivation
repetition quality is more important than practice duration
“talent” mattered far less than I assumed

This made me wonder how many domains feel like talent, when they are simply starved of feedback or structure.

If you want to try it

Here is the iOS link:
[https://apps.apple.com/app/pi-deliberate/id6757496426)

If you try it, I’m interested in:

how many digits you reach
when chunking starts to appear (if at all)
whether the loop feels rewarding or exhausting

Data is more interesting than download numbers in this case.

A question for IH

Has anyone here worked on:

deliberate practice
memory training
spaced repetition
cognitive architecture
learning apps
educational UX

I’m curious about the expected “ceiling” before diminishing returns. The Guinness record is 70,000 digits (Rajveer Meena, 2015), which raises questions about what the real bottleneck is.

Closing thoughts

This experiment pushed me toward the idea that talent is often just feedback loop design and time allocation. Once the loop exists, the experience becomes less about willpower and more about tuning a system.

If anyone has experience or resources in this area, I would appreciate pointers.