How many hours of deep work a day, and how to measure yours
There is no single number of deep work hours that research hands you, and a better starting point is to measure what you actually do for two weeks. The useful question is then how your own number moves, not whether it matches someone else's. Below is what deep work means, what the evidence does and does not say, and how Hina 0.6.0 counts it.
What deep work means
Cal Newport introduced the term in his 2016 book Deep Work. He describes it as professional activity performed in a state of distraction-free concentration that pushes your cognitive capabilities to their limit; the wording varies slightly between quotations, but that is the core. He contrasts it with shallow work, the logistical, lower-demand tasks that are often done while distracted.
The definition is about the quality of attention, not the type of app. Writing a design document in a quiet room is deep work and the same document written between chat replies is not, even though the editor was open the whole time. That distinction matters later, because a computer can see the app and the timing but not the attention.
What research says about daily hours
The study people usually cite is Ericsson, Krampe and Tesch-Romer's 1993 paper on deliberate practice, which compared student violinists at different skill levels using retrospective estimates and a week-long practice diary. It reported estimated typical weekly hours, about 29.8 hours a week of practice alone for the best and good groups combined, against 13.4 hours for the less accomplished group, and it describes the top performers as keeping practice at a high level without it leading to exhaustion. Those are weekly estimates from student violinists, not a daily prescription for office work.
The four-hour figure people quote does come from this paper, but as a hypothesis. The authors note that training studies found essentially no benefit from practice beyond 4 hours a day and reduced benefits beyond 2, and that their expert violinists and pianists did around 4 hours of deliberate practice a day, which they offer as the possible general limit on practice that can be sustained without exhaustion. That is a limit on effortful practice in music and sport, inferred from a small number of studies, not a measured ceiling for knowledge work.
It also needs to be read with care. A 2019 replication by Macnamara and Maitra reported that they did not replicate the core finding that accumulated deliberate practice corresponded to each skill level, and that practice alone explained 26 percent of the variance in performance against 48 percent in the original. Violin practice is also a different job from writing code or a report.
So the supportable summary is modest. Concentrated effort has a ceiling that differs by person and by day, and none of the studies above gives three or four hours as the right amount for your job. That is why measuring your own baseline is more useful than adopting a target.
How Hina counts a deep work block
Hina records the foreground app every 3 seconds and treats 180 seconds without input as idle. Each app is assigned a category, and categories belong to three groups. Coding, Writing, Design and Meetings are in the Work group; Browsing, Comms, Reading, Utilities and Uncategorized are Neutral; Media is Leisure. Deep work is built only from Work group time.
A block starts at the first Work session and keeps going while Work sessions follow each other. A gap of up to 5 minutes does not end it, so a quick look at chat in the middle of a task is absorbed. A longer gap of up to 15 minutes also does not end it if at least 80 percent of that gap was idle, which covers waiting on a build or an AI answer, thinking away from the keyboard, or stepping away briefly. A block counts as deep work only when its real work time reaches 10 minutes; the app's own empty state reads No stretch of deep work over 10 minutes yet.
Two consequences are worth knowing. Meetings are in the Work group, so a day of calls can show a lot of deep work that was not solitary concentration, and Reading is Neutral, so an hour with a paper open does not count. The number is a description of continuous work-type activity, and the sibling article on understanding focus time walks through the edge cases in more detail.
Reading the weekly numbers
The Week tab shows Active this week, Deep work and Daily average, with your longest day under the average and a comparison to last week in the summary that appears when you hover that row. Its Golden hours card names the start hour that accounts for the most deep work. The Month view adds Median deep work, which is steadier than an average because one unusual day does not move it much, and a Deep work by weekday card with the average deep work for each day of the week.
The Interruptions card counts a different thing: a break of under 5 minutes in the middle of work, attributed to the app or site you went to. A trip out that comes back to the same app within 3 minutes, such as waiting on an AI reply, counts as a pause rather than an interruption. A day can have long deep blocks and many interruptions, or few of both, and the two numbers answer different questions.
Treat the first week as a baseline and ignore the comparison with last week. A holiday, a release day or a week of meetings will move the number more than any change in habit.
Raising it slowly
Start from your median, not your best day. If the Month view says your median is 70 minutes, aim for 80 or 90 on a few days, not three hours on all of them. Look at what the Golden hours card says and try putting your hardest task at that hour before changing anything else.
Then look at the Interruptions card. If the biggest source is one chat app or one site, grouping those visits into a set time is cheaper than trying to resist each one. Remember that the 5-minute tolerance hides short looks from the deep work number but not from your attention, so a block with five interruptions in it is not the same as an uninterrupted one even if both are counted.
Do not read a low number as a verdict. Hina does not see thinking on paper, reading a printed page, a whiteboard session or a walk. If the idle review card asks about a stretch of 10 minutes to 2 hours with no movement, answering Reading / thinking classifies it as Reading, which is Neutral and still does not add to deep work. Only Work activity does, so the number understates offline concentration by design.
Where the data lives
Hina stores sessions in a SQLite file at ~/Library/Application Support/Hina/hina.sqlite, takes no screenshots, needs only the Accessibility permission, and records input as counts, never content. The app reaches the network in three places only, each on a click and none carrying activity data: the checkout page, activating a purchased key, and Check for Updates. If you would rather pull numbers into your own spreadsheet, Settings has an Export CSV action.