The short answer
Some knowledge workers show a late-evening activity peak in telemetry—especially in certain remote/hybrid contexts. That pattern is useful for questioning “9-to-5 only” assumptions; it is not proof that late-night coding is healthy, equitable, or optimal for you.
What the study describes (high level)
The work analyzes anonymized keyboard activity to characterize daily rhythms, reporting an emerging third peak for some participants. Remote days showed differences relative to onsite/hybrid patterns in aspects of the findings—exact effect sizes and subgroup details belong in the paper, not a vendor blog paraphrase.
The core developer takeaway is modest: work rhythms are not monolithic, and hybrid schedules may redistribute activity across the day.
What it does not mean
It does not mean “everyone has a third peak,” that night work is productivity-maximal, or that managers should schedule work at night. Telemetry shows behavior—not wellbeing, quality of decisions, or sustainable pace.
Developer translation
If your honest chronotype and constraints place strong focus late, you may choose to protect that block—while guarding sleep. If your peak is morning, ignore internet narratives that glamorize midnight shipping.
For placement of maker blocks around meetings and meals, see sprint timing and meal choreography.
Chronotype and equity
Global teams already distribute “night work” unevenly. Using triple-peak language to normalize after-hours availability can harm caregivers and timezone-distant colleagues. Policy should not outsource equity to individual grit.
Pairing with sprint design
Whatever your peaks, sprint design still needs explicit start/stop, instrumental audio policy if you use music, and session proof—so “I worked late” does not devolve into unbounded thrash.
Methodological humility
Keyboard activity is a proxy for work. It does not measure thinking in the shower, reading on paper, or mentoring without typing. It also does not measure harm: late-night work can show up as productivity while increasing burnout risk.
When vendors cite academic papers in marketing, ask what was measured, in which population, and whether the vendor’s product appears in the study. The answer is frequently “no.”
Practical takeaway
Triple-peak days are a useful reminder that activity curves vary—especially in hybrid work. They are not a license to glorify night coding. Pair telemetry narratives with sleep, equity, and quality signals your team actually cares about.
Frequently asked questions
Should I schedule coding sprints at 9pm because of the third peak?
Not automatically. A third peak in telemetry is not a prescription—especially if sleep debt and stress rise. Use it as a prompt to examine schedule shape, not as permission for unsustainable hours.
Is this the same as circadian rhythm advice?
Related but not identical. Triple-peak language comes from observed keyboard activity patterns in a specific dataset—not a universal biological law.