Deliberate Practice for Painters

The phrase "10,000 hours" gets thrown around a lot. Most people who use it haven't read Anders Ericsson, whose research the number is loosely drawn from. The book is Peak (2016), and its central argument is much narrower than the cliché. It's not that 10,000 hours of anything produces mastery. It's that 10,000 hours of a very specific kind of practice — deliberate practice — produces mastery, and most of the hours people log in any given pursuit don't qualify.

This is worth taking seriously if you paint, because it explains something most working painters notice eventually: a lot of painting time doesn't seem to make you better. You can do it for years and stay at the same level.

The cause is almost always that the practice isn't deliberate.

What Ericsson actually said

Three conditions, roughly:

A clearly identified target skill. Not "get better at painting." Something narrower — "improve the value transitions in skin tones," "lay in a grisaille faster," "match Sargent's edge quality in a head study."

Immediate, specific feedback. Not "this looks nice." Specific external comparison: against the model, against a master copy reference, against a teacher's critique, against your own previous best attempt.

Operating at the edge of current ability. Not in your comfort zone. The practice has to be slightly beyond what you can already do, which means most of your attempts will fail, which is uncomfortable, which is why most people drift back into doing what they already know how to do.

If you spend an evening at the easel painting something pleasant and finishing it without ever identifying a specific weakness you're working on, comparing your result to a known standard, or attempting something you can't quite do — you've practiced. You've put in the hours. But it doesn't really count for deliberate-practice purposes. Ericsson would say you've practiced staying the same.

Why painting is structurally hard to practice deliberately

Most domains where deliberate practice works well have a clear external standard. A chess move is right or wrong. A free throw goes in or it doesn't. A scale is in tune or it isn't.

Painting doesn't have this. The standard is the painting you intend to make, which only exists in your head, which makes it easy to redefine after the fact. You can spend three hours on a passage and decide retroactively that the result is what you meant to do. Most painters do this constantly without realizing it. It's the path of least resistance and it kills practice.

The atelier tradition solved this by introducing external standards aggressively. The Bargue plate is the external standard for early drawing. The plaster cast is the external standard for value and proportion. The master copy is the external standard for the full painting procedure. You're not painting something — you're painting that specific thing, and you can be compared against it.

This is the structural reason master copy keeps coming up. It's not because copying is intrinsically valuable. It's because it's one of the few painting activities that comes with a built-in external standard you can't redefine after the fact.

What we tried to bake into the recreation guides

When we built our 2,491 grounded recreation guides, we made a few choices that I think push the practice toward the deliberate end of the spectrum:

The unit is a specific painting. Not "Vermeer-style" — A Lady Writing. The target is unambiguous.

Each guide has a phase structure with explicit transitions. Underdrawing → underpainting → first pass → refining → finishing → varnishing. You can check whether you're at the phase you think you're at.

Most steps cite a source. If you disagree with a step, you can go read what the source actually says and form your own view. This sounds small. It's the difference between taking instruction and participating in a practice.

The knownGaps block names what the corpus doesn't cover. The guide isn't pretending to be complete. There's room for you to be the one who notices things the guide didn't.

What we couldn't put in: the external corrector. A master copy without a teacher critiquing it is missing the most important feedback loop. We don't have a good answer for that yet. The best workaround we know of is photographing your work-in-progress next to the original, at the same size, in the same lighting, and looking at the two side by side without flinching. It's not as good as a teacher. It's better than nothing.

A deliberate-practice session, concretely

Rough shape, if you wanted to try:

1. Pick a specific weakness from your last painting. Not a general one. A specific one. My midtone transitions in skin are muddy. My horizon lines wander. My darkest darks read green when they should read warm.
2. Find a passage in the collection — could be one square inch of a Sargent head, one small Constable sky — that demonstrates the thing you want to fix.
3. Set up to paint that passage at the same size as the original, with the original visible next to your easel.
4. Paint it. Compare. Don't redefine what you were trying to do. The original is right there.
5. Paint it again, with the corrections you can see from the first attempt.

That's a 90-minute session. It's much less satisfying than painting a complete new picture. It also produces more improvement per hour than almost anything else you can do at the easel.

The classical atelier tradition has been quietly running this loop for two centuries. We're trying to make the external standards part — the specific painting, the source-cited procedure — available to people who can't get to an atelier.

The feedback loop is on you. There's no version of this where someone else does that part.