01 / The word
Arete. ἀρετή.
One old word, kept because it names the standard the work is held to. What follows is where the word came from, and what the research behind the idea actually says, including the parts that did not hold up.
What the Greeks meant by it
Arete is an old Greek word. It is usually translated as excellence, but that is a little grand for what it meant.
A knife has arete when it cuts. A horse has arete when it runs. A person has arete when they do their own work as well as it can be done. It is not a compliment about who you are. It is a description of a thing doing what it is for.
So it was measured in practice, not in ambition. You had it on the days you did the work properly. You lost it on the days you did not.
02 / The definition
Aristotle files it as a state, not a feeling
Aristotle spends Book II of the Nicomachean Ethics on this. He asks a plain question: what kind of thing is excellence. A feeling. An ability. Or something else.
He rules out the first two. We are not praised or blamed just for feeling something, so it is not a feeling. We are not called good or bad by birth, so it is not a natural ability. What is left is a hexis, which W. D. Ross translates as a state of character. A settled way of being that you built.
And you build it the way you build any skill. In Book II, chapter 1, he writes that men become builders by building and lyre players by playing the lyre.
So too we become just by doing just acts, temperate by doing temperate acts, brave by doing brave acts.
The doing comes first. The quality comes after. The word already carried the whole claim.
03 / The whole life
One good day is not the measure
Earlier in the same work, in Book I, chapter 7, Aristotle adds a line most people know without knowing where it came from. One swallow does not make a summer, nor does one day.
He is writing about a happy life there, not about skill, so I am borrowing the shape of the argument rather than the subject. But the shape holds. A short run of good work settles nothing. The measure is what you do across a long stretch of time.
This is the part I keep coming back to. Most business advice is written about the good day: the good launch, the good quarter, the good hire. Arete is written about the average day.
04 / The modern version
The research that tried to prove it
More than two thousand years later, a psychologist called Anders Ericsson made a similar argument with data. In 1993, with Ralf Krampe and Clemens Tesch-Römer, he studied violin students at a Berlin music academy.
The paper described something it called deliberate practice. Not just doing the thing. Doing it under conditions the paper set out.
What the 1993 paper asked for
- A task with a clear point.
- A difficulty set just past what you can already do.
- Immediate, informative feedback.
- The chance to repeat it and correct the errors.
By age twenty, the best violinists had accumulated around ten thousand hours. Ericsson later described that figure as an average for the group, not a threshold anyone had to cross. It left the paper anyway and became a rule.
Ericsson spent years objecting to it. Writing with Robert Pool in 2016, he said there is nothing special or magical about ten thousand hours, that the round figure was Malcolm Gladwell's choice because it read well, and that half of the ten violinists in the top group had not actually reached it by that age.
05 / The correction
Then it got tested, and it got smaller
In 2014, Brooke Macnamara, David Hambrick and Frederick Oswald pooled 88 studies covering 11,135 people. They asked how much of the difference in how well people perform sits with how much they had practised. The answer changed sharply by field.
| Field | Explained by practice | Not explained |
|---|---|---|
| Games | 26% | 74% not explained |
| Music | 21% | 79% not explained |
| Sports | 18% | 82% not explained |
| Education | 4% | 96% not explained |
| Professions | under 1% | over 99% not explained |
In 2019, Macnamara and Megha Maitra rebuilt the original violin study, this time with a double blind procedure. They did not replicate the core finding.
Practice alone up to age eighteen came out at about 8,224 hours for the best violinists and about 9,844 hours for the good ones. The difference was not statistically significant, and the group with more hours was not the better group. Across all three skill levels, practice accounted for 26 per cent of the difference in performance, against 48 per cent in the original study.
I am putting this on my own page because it argues against the easy version of what I sell. If the claim were only that hours produce excellence, the evidence would be thin, and thinnest of all in the professions, which is where my clients work.
06 / What held up
The two findings that did replicate
Strip out the hours and two nearby findings are much steadier.
Spacing
Nicholas Cepeda and colleagues pooled 839 measurements from 317 experiments in 2006. Across them, spaced sessions averaged 47.3 per cent recall against 36.7 per cent for the same material done all at once. Same effort, different spread, better result.
Retrieval
Henry Roediger and Jeffrey Karpicke, also in 2006, tested reading something again against being asked to recall it. After five minutes, rereading won. After a week, recall won by a wide margin. Their words: prior testing produced substantially greater retention than studying.
Both are laboratory studies about remembering words and passages. Neither is a study of running a company, and I am not going to stretch them into one.
But the shape is worth having. Work spread out and pulled back out again beats work crammed in and read over. And the thing that looks worse in the first five minutes is the thing that is still there in a week.
07 / The loop
Feedback is not good just because it is feedback
The method has a loop in it. Finished work goes back into the memory, and the next pass starts from what the last one learned. That loop is the part of the method the word arete is actually pointing at.
There is a warning attached to it, and it is a serious one. In 1996, Avraham Kluger and Angelo DeNisi pulled together 131 papers, 607 measured effects and 12,652 people. On average, feedback helped: the weighted average effect was 0.41 on the scale they used. But over 38 per cent of the measured effects were negative. Their abstract puts it in one line: over a third of the feedback interventions decreased performance.
Their explanation is about where attention lands. Feedback that points at the task tends to help. Feedback that points at the person tends to hurt, because attention moves off the work and onto the self, and the work gets less of you.
work done → what happened, written down → into the memory → next pass starts there
So a loop is not good because it is a loop. It is good when the thing coming back is about the work. That is a design rule, not a slogan. It is why the weekly review in the method reads the output and the numbers and reports on those, and why it never grades you.
08 / How long it takes
Sixty six days, and a very wide range
66
range: 18 to 254 days
People want a number for how long a new habit takes. There is one, and it is less tidy than the version you have heard.
In 2010, Phillippa Lally and colleagues at University College London followed 96 volunteers. Each picked one daily behaviour and repeated it in the same context, after breakfast for example, for twelve weeks. The median time to reach a plateau was 66 days. The range ran from 18 days to 254.
Two honest notes. Of the 82 volunteers whose data could be used, only 39 produced a curve that fitted the model well, so the 66 day median describes those 39 and not everybody. And the study ran for 84 days, so the far end of the range is a projection rather than something anyone watched happen.
The finding I care about most is the small one at the end. Missing one opportunity to perform the behaviour did not materially affect the habit formation process. One missed day is not a broken run. That is worth knowing on the Tuesday you skip.
09 / The name
Why the method carries it
The method is called Arete AIOS (AI Operating System). ChatGPT and Claude provide the engine. The method builds and keeps the business brain that engine works from: what your company knows, the work you repeat written down once, and the tools it is allowed to touch.
The word sits on it as a limit, not as a promise. It says the goal is the repeated work done properly, not more output for its own sake. It says the judgement calls stay with you, because arete was never a property of a tool. It was a property of a person doing their own work well, and a tool cannot hold it on your behalf.
If the method were called Boost or Flow, the name would be promising you a feeling. Arete names a standard, and a standard is a thing you can fail.
Written by Egils Boitmanis, who runs his own business on the same system he teaches.
The morning brief, the numbers and the weekly review come out of his own files.
About Egils10 / The standard
What this page will not say
It will not say that a system makes you excellent. The research further up is exactly why. In the professions, which is the field my clients are in, how much people had practised explained under 1 per cent of the differences in how well they performed. Hours are not the mechanism.
What I will say is narrower, and I think it survives the evidence. The work you repeat can be written down once. What your business already knows can sit in one place instead of being retyped into an empty window every morning. Finished work can go back in, so the next pass starts further along than the last one did.
That is a loop with a short cycle, a written record, and feedback aimed at the work. Spacing, retrieval and feedback are the three findings above that held, and all three are about that shape.
The rest is you, doing the work on the average day. That was the deal in Athens too.
Where these come from
- Aristotle, Nicomachean Ethics, Books I and II, W. D. Ross translation
- Ericsson, Krampe and Tesch-Römer, Psychological Review, 1993
- Ericsson and Pool, Salon, 2016
- Macnamara, Hambrick and Oswald, Psychological Science, 2014
- Macnamara and Maitra, Royal Society Open Science, 2019
- Kluger and DeNisi, Psychological Bulletin, 1996
- Cepeda, Pashler, Vul, Wixted and Rohrer, Psychological Bulletin, 2006
- Roediger and Karpicke, Psychological Science, 2006
- Lally, van Jaarsveld, Potts and Wardle, European Journal of Social Psychology, 2010
All sources checked on 21 August 2026.
11 / Where to go next
Free. Nine questions, about four minutes. Your first role shows right away.