You did the “right” evening. Closed the laptop on time. No drama. No 2 a.m. doomscroll. And still your sleep comes out thin. 2–4 wake-ups. Or it ends early with that instant mental boot, like a machine that technically went to sleep but kept a few background services running.

If that sounds familiar, this article is here for one specific idea. Thin sleep is often not a bedtime failure. It is a daytime system problem. The brain can stay in a kind of monitoring mode across the full 24 hours, and remote work makes that easier to trigger because there are fewer natural off switches. The result is not always “stress” in the dramatic sense. It is smaller desk behaviors that pile up and keep the nervous system on watch.

What you will get here is pragmatic, low-ceremony ways to reduce that monitoring loop without adding another app, plan, or perfect routine. We will look at:

The lens is intentionally a bit systems-minded. The author is french, born in 1974, physics-trained, and has spent most of adult life at a desk across Beijing, Berlin, and now Lisbon, with a strong bias toward metrics and skepticism toward polished sleep tips that ignore the work system generating the churn. If advice that starts at 22:30 has been useless for you, good. That probably means you are not broken. It just means you are debugging the wrong layer.

Thin sleep is often a daytime problem

Thin sleep even when evenings are calm

The useful question is what your next morning looks like.

A lot of insomnia works like this: the “on” state doesn’t only live at bedtime. It can linger all day and then leak into the night (Riemann et al., 2015). And fragmented sleep can feel unrecovering even when total time looks decent on paper (Stepanski, 2002). Desk-level implication: if you wake up and immediately start scanning for what changed, your workday is already feeding the loop you’re trying to stop at night.

At the desk, you can see the logs. Not dramatic symptoms. Just small behaviors that pile up.

It is tempting to blame the usual bedtime suspects. After-hours pressure is real. Telepressure can hurt recovery and sleep (Barber & Santuzzi, 2015). Phone work intrusions can also block detachment (Derks & Bakker, 2014). But thin sleep also shows up after a disciplined evening, because the volatility happened earlier and trained the brain to keep polling for changes.

Work that keeps you on watch

Your brain stays in polling mode

Remote work makes this pattern easier to trigger because the day has fewer forced handoffs. Work starts to feel like light incident response. Not the sirens, just the constant monitoring.

Priorities flip. Decisions reopen. A stakeholder drops a “small comment” and suddenly the plan is back in draft. When things feel unpredictable and out of your control, the body often reacts more than people think, including cortisol responses (Dickerson & Kemeny, 2004). If insomnia is often a 24-hour arousal problem, it makes sense this leaks into the night (Riemann et al., 2015).

Remote work removes the off switches

In remote work there is no commute buffer, fewer corridor handoffs, and everything stays editable, reachable, and “quick to fix” from the same chair on the same screen. Even “quiet” time gets perforated by pings, doc edits, and drive-by questions.

Flexibility can quietly turn into higher expectations of responsiveness, not less (Mazmanian, Orlikowski & Yates, 2013). When the environment never changes, the brain gets fewer cues that something is truly finished.

A metrics lens makes the mismatch obvious

I rely a lot on metrics, including sleep tracking, which makes me skeptical of polished sleep tips that ignore the work system generating the churn.

If the day trains “stay available, everything can change,” advice that begins at 22:30 tends to treat the symptom, not the cause.

Priority volatility trains monitoring

Volatility is not drama but it is a signal

Priority volatility is when “top priority” changes after lunch, a stakeholder comment reopens what was “done,” and quick pivots get rewarded while completion stays negotiable. At the desk, this means more interruptions and more switching.

Interruptions reliably raise stress and workload in knowledge work (Mark, Gudith & Klocke, 2008). Every switch has a resumption lag that makes closure harder (Altmann & Trafton, 2002).

Here’s the annoying part:

Unstable work states teach the brain to check more often.

Open loops follow you to bed

An open loop is unfinished work that still feels active because there is no stable end state yet. Late changes are perfect at creating these loops, and they keep popping back up as mental replay.

A basic plan can reduce intrusive thoughts from unfulfilled goals (Masicampo & Baumeister, 2011). Without that, work rumination is linked to poorer sleep quality (Querstret & Cropley, 2012). So even with a calm evening, bedtime becomes mental drafting—and the thought that keeps tapping you on the shoulder is: “If they comment tomorrow, I’ll have to reopen it again.”

Unpredictability has a body level cost

Unpredictable or uncontrollable stressors reliably activate stress responses, including cortisol (Dickerson & Kemeny, 2004). Also worth saying out loud, few studies isolate “priority volatility alone” as a clean causal variable. But the stress and rumination pathway it triggers is pretty well supported, especially in familiar desk patterns like reopen cycles after 16:00 and acceptance criteria that keep shifting.

A 2 minute volatility check that does not become another app

Two questions

1) How many times did the top priority change after 15:00 (a new “top 1” item, not just a subtask)?

Late flips are expensive because they steal the last chance for closure, and you go to bed still building a plan.

2) How often did “done” become “not done” after someone saw it?

Call it a reopen cycle, not a personal failure. This is exactly the kind of unfinished business that tends to fuel rumination, which is linked to poorer sleep quality (Querstret & Cropley, 2012).

A quick tell is what the mind replays at night. If it is mostly dependencies, updates, and “what if they comment and it all reopens,” volatility is a strong suspect.

Pick 1 main source of churn

Think of something like a doc that gets reopened with “small” edits at 17:20, again, after you already tried to land it.

Picking only 1 matters because “fix everything” becomes yet another project.

Build a small firewall against late day churn

A freeze window that keeps the afternoon finishable

After a fixed time like 15:30, work can still execute and close, but it does not expand in scope unless it is truly urgent. If a new request lands at 16:10, it gets queued.

I like boring rules because my brain loves exceptions.

It’s not magic:

A freeze window only works socially if “urgent” has an actual definition and a path. One useful setup is a real escalation channel for emergencies, and a calm response-time norm for everything else. Flexibility often increases expectations of responsiveness (Mazmanian, Orlikowski & Yates, 2013). Predictability beats vague availability.

Make done harder to reopen with 1 sentence

Attach a 1-sentence definition of done to the deliverable. What “done” means, what is out of scope, and what would count as a legit reopen trigger.

Here is a template, because nobody wants another doc.

Done means it meets agreed acceptance criteria and quality standards, is verified or tested, ready to release or operate with stakeholder acceptance.

This reduces sleep friction because closure becomes more believable. Less “tomorrow they will invalidate today,” more trusting state transitions.

Put open threads on the calendar with next touch times

A small rule is that every open thread gets a specific next touch time instead of “later.” It turns vague waiting into a plan. That tends to cut down on intrusive thoughts.

Examples that sound like a real Tuesday.

Scripts that reduce churn without sounding difficult

Park it and name a real next touch

For new pings, be responsive but not instantly interruptible.

“Seen. Parking this for tomorrow 11:00 so i can finish the current block.”

For peers or leaders.

“I can take this now, but then X moves to tomorrow. OK?”

This only works if tomorrow 11:00 actually happens. A fake next touch trains the brain to keep checking anyway.

Tradeoffs that force clarity and stop reopen loops

When scope tries to expand late.

“If this changes scope, which item drops or moves?”

To reduce reopen cycles.

“Happy to revise. What is good enough here, in 2 bullets?”

To calm async whiplash.

“Can we batch feedback in 1 message by 14:00 so i can close this today?”

A 3 day test for relief not perfection

Day 1 is a tiny baseline

If days are packed and sleep is already thin, the last thing needed is a tracking project. Do 1 baseline day with a paper sticky note and log only 3 items, nothing else.

1) priority changes after 15:00

2) reopen events after “done”

3) morning scan urge before starting work

Days 2 and 3 use only 2 tools

For days 2–3, change just 2 things.

Example combos.

Keep it minimal. More rules can become extra load.

What working looks like

Success is boring and practical. It is less monitoring, not perfect sleep scores.

If none of this shifts after a few tries, it is ok to stop debugging the work system and look for proper clinical help.

When it is time to get real help

This piece is about common desk-work patterns that keep the brain in monitoring mode, not medical advice.

Get evaluated if there is dangerous daytime sleepiness, near-miss driving moments, loud snoring or choking or gasping, breathing pauses noticed by others, or high blood pressure with unrefreshing sleep. For sleep apnea, symptom-based evaluation matters, and questionnaires are not diagnostic (AASM, 2017).

For chronic insomnia, guidelines recommend CBT‑I as a first-line option (ACP, 2016; AASM, 2021). In plain terms:

Meta-analyses support meaningful improvements for many people, without promising miracles (Trauer et al., 2015; van Straten et al., 2018).

If the workday test helps, keep the smallest rule and delete the rest. If it does nothing, that is not a character flaw. It just means the main driver is elsewhere.

Sleep should do recovery, not change control.

What part of the day keeps reactivating your brain the most, late scope creep, unclear ownership, or the “just in case” checking?