Anders Ericsson & deliberate practice

Source: https://influences.sgit.ai/register/ericsson/index.html↗ · markdown twin of the entry page.

An influence on Dinis Cruz — one of 25 entries in his register.

• tier traced — corpus evidence exists today
• kind person + research
• status full — the seven-block register format
• briefing none

The research behind the ten-thousand-hours number, attributed to the person who did it rather than the person who popularised it — a distinction the corpus is careful about, which is itself the tell.

Block 1 — The anchor

Peak: Secrets from the New Science of Expertise — K. Anders Ericsson and Robert Pool — 2016

The book is the accessible anchor; the 1993 paper The Role of Deliberate Practice in the Acquisition of Expert Performance is the primary source and is in Block 7.

Linked, never rehosted.

Block 2 — In his own words

Two things resonate here, and the second is the one that matters now. The first is the correction itself — that experience and practice are different things, and that twenty years of doing something the same way is not twenty years of practice. Attributing the research to Ericsson rather than to its popular summary is a small act of the same discipline this whole site runs on: cite the claim's actual source, and the claim gets sharper.

The second is what deliberate practice means when a model writes most of the code. If expertise requires working at the edge of your ability with immediate feedback, and generation removes the part where you struggled, then the practice loop has been cut — not made faster. That is an uncomfortable implication for an estate built on AI-assisted development, and it belongs on this site rather than being left implicit.

The overlap with flow ↗ is not accidental: challenge matched to skill is Csikszentmihalyi's third condition and Ericsson's core mechanism, described by two research programmes that were not talking to each other.

Block 3 — The principle

Expertise comes from structured practice with immediate feedback at the edge of current ability — not from hours, and not from experience.

The number is the least interesting part and the only part most people repeat. Ericsson's actual finding is about the structure of the practice: a specific target, immediate correction, and deliberate work on what you are worst at.

Block 4 — The trace table

No trace table yet. The mining run found the volume and the attribution discipline; it did not record which documents make which argument. This entry's real trace question is sharper than a path list, and worth naming while the table is empty: does anything in the estate's methodology preserve a practice loop, or does it only preserve a delivery loop? R2 ↗.

Block 5 — The gaps, as build specs

G1 — Nothing in the methodology is designed to make someone better

The estate's development methodology optimises for flow and throughput. Deliberate practice requires the opposite of flow in one specific respect: working at the edge, where you fail. A build spec that would take this influence seriously: a mode in which the model explains rather than generates, deliberately chosen for the parts the person is weakest at. Nothing like it exists, and the entry says so rather than claiming the methodology covers it.

Block 6 — The checklist

• Is this practice, or is it repetition? What is being worked on that was not already easy?
• Where is the immediate correction? Practice without feedback is just time passing.
• Is the difficulty at the edge of ability, or comfortably inside it?
• If a model does the hard part, who is getting better — and does that matter here?

Block 7 — The wider library

• The Role of Deliberate Practice in the Acquisition of Expert Performance (1993) — the primary source, and the paper the ten-thousand-hours number was extracted from https://psycnet.apa.org/record/1993-40718-001 ↗
• Peak (2016) — Ericsson's own correction of the popularisation, written because the summary would not stop spreading

The corpus evidence

CC BY 4.0 — Dinis Cruz, with AI co-authorship (Claude, Anthropic). The anchor work belongs to its author and is linked, not licensed here.

source: /register/popper/index.md