Friday evening. You sit down to write the week's recap and the recap won't come. You did things — a lot of things. New tools tried, articles half-read, three content formats started, two experiments thought about. Nothing went deep. Nothing finished. The week feels like static under your hands: busy, full, and somehow less than the sum of its hours.
Maybe scattered isn't the cost of curiosity. Maybe it's the cost of mixing two modes — exploring (trying new things) and exploiting (going deep on one chosen thing) — without the small fence that lets both breathe.
That made me think about the explore/exploit week experiment.
The question: What changes when I separate exploration from exploitation each day?
The hypothesis: if I keep a small explore block and a longer exploit block in each day for seven days, separated by the clock, I'll feel less scattered and make clearer commitments. Not because curiosity got smaller. Because curiosity now has a fence, and so does depth.
The signal: by day seven, I can name one thread worth exploiting for the next 2–4 weeks, and I've protected at least five exploitation blocks from being eaten by novelty.
What you do for 7 days
- Explore block: 20–30 minutes. Try one new idea, skill, format, tool, or angle. Allowed to be messy. Allowed to go nowhere.
- Exploit block: 60–90 minutes. Go deep on one chosen thread. Same thread every day, all week, if possible.
- The fence. Exploration cannot eat exploitation time. If a shiny thing appears in the exploit window, name it on a list ("for next week's explore") and return to the thread.
- Daily capture, one line.
- Week review (day 7). What thread wants a 2–4 week exploitation season? What exploration inputs were high-signal vs. noise?
What this experiment grows
Not focus. The muscle of finishing what you started, without giving up the curiosity that made you start. Most one-person businesses don't lack discipline; they lack a way to honor both modes at once. The fence does that. It isn't a moral judgment — it's just a clock.
After seven days, the value often isn't only in the exploitation thread that emerged. It's also in the cleaner explore time. Twenty minutes of explore, fenced, produces more usable signal than three hours of explore that bled into everything else.
Where it pairs
This experiment walks with the afternoon swamp — the hours where the day decays into half-tasks and small avoidances because nothing has a clean shape. The fence gives the afternoon a shape. A daily routine with an AI assistant — the Inbox-to-Decision workflow, here a workflow with AI — pairs naturally: it sorts what belongs in each block, so the explore window stays explore and the exploit window stays exploit.
One week. One small fence. The thread, finally chosen.