When you break down the day of someone who is "slow at work," they actually have very little time where their hands are stopped.
Making documents, replying to emails—the hands typing on the keyboard are probably fast. The delay happens after an interruption from Slack or other notifications.
In other words, it takes a long time to return to the original screen and remember how far you got. I think the difference lies in the timing of "restarting."
Being told "do it faster" only speeds up the work itself, so the gap doesn't narrow much.
I think this is a major trap when trying to handle multitasking as single-tasking.
A research team from UC Irvine observed 24 people working in an office for three and a half days. It took an average of 25 minutes for individuals to return to interrupted tasks. Moreover, during that time, they had started two or more other jobs.
So, what's being eroded isn't "hand speed," but those 25 minutes to return. People who are interrupted many times a day just accumulate these 25-minute blocks; their typing speed is sufficient.
I still take time to "restart."
Since I run several Claude Code sessions simultaneously, my day is full of interruptions. But the actual typing time is short. What's long is returning from the switch and remembering how far I progressed. Especially, I feel that writing tasks disrupted by AI interactions suffer from reduced concentration.
Productivity won't improve unless we make "brain switching," not just "task switching," more rational.
Slack notification pops up -> check another thing while replying -> return to original screen -> Claude session finished, give next instruction -> return to original screen -> remember how far I got -> finally hands move...
The truth behind "being slow at work" is not hand speed, but "time until restart."
I don't think work gets faster without solving this.
So, instead of speeding up my hands, I focus on "how to return," using "three rules" to compensate.
1. Place interrupted requests in a designated spot before returning to original work
-> If you keep a "please confirm" from Slack in your head, you waste brainpower trying not to forget. So, I use Trello (task management tool) like a cockpit.
Two sections: a column with "Today's Tasks (Long Tasks)" sorted by priority, and a box for "Tasks for Gaps (Short Tasks)" with sticky notes. Incoming requests go to the gap box if quick, or inserted into the priority column if time-consuming.
Actual task management view in Trello
Every time a task finishes, I look at this screen, so placed requests aren't forgotten. When returning, only the original task remains in my head.
2. Don't stop at natural breaks. Intentionally stop mid-way when taking a break
-> I think this is the most important point. A common mistake is finishing "up to a good stopping point" before lunch or a 15-minute break. This causes the startup cost and concentration ramp-up to overlap upon return. Stopping mid-way is best. If you stop halfway, you resume from the "continuation," reducing startup load.
3. Write down the first 5-minute task for tomorrow one hour before leaving, and start with it in the morning
-> This is the daily version of rule #1. The next morning starts with "yesterday's continuation" rather than zero. Opening notifications before starting anything means spending the whole morning "in the process of returning." Open emails and Slack only after your hands have moved once.
My solution to "being slow at work"
"Please confirm," "Urgent"
In workplaces where these fly in multiple times a day, especially for those managing multiple processes alone, the cause of slowness is often thought to be "the number of notifications."
However, there's more to the UC Irvine observation. For analytical and development roles, interruptions were roughly half caused by others and half self-imposed. Even without calls or notifications, people stop themselves. Moreover, the time to return was longer for self-interrupted tasks than for externally interrupted ones.
What needs to be sped up is the subsequent "restart," not the hands. The three rules above are my systems for signaling myself to return.
Try intentionally stopping at a "bad breaking point" before your next break.
I also explain timely topics and hints for tomorrow's work structurally and clearly.
https://x.com/smark_x/status/2098393324768157849
https://x.com/smark_x/status/2096168360103088156
- Overthinking prevents action
- Perfectionism leads to running out of time
- Wanting to blame lack of sense when hitting walls
- Work hasn't gotten easier despite adding AI
I suffered from all these but made work easier by systematizing with "Logical Thinking." I post daily, so please follow if interested. → @smark_x
