It’s 21:30 and the workday is technically over. But the laptop stays nearby. Dinner takes longer than it should. Your phone is face up, within reach, not because you’re scrolling, but because you might need to respond. Not a lot is happening. That’s the problem.

For a lot of desk jobs, the hardest part about sleep is not falling asleep. It’s that light, on-standby sleep where you’re off, but not really off. A message could land. A thread could move. And even if nothing happens, your brain keeps doing tiny background checks anyway. Like a system waiting for an interrupt.

This article is about that pattern and what to do with it without turning your evenings into another project. You’ll get a clear explanation of why random after-hours reachability hits recovery so hard, including telepressure (that internal “reply now” itch) and intermittent reinforcement (random rewards that train checking). You’ll see what “standby sleep” looks like at night and the next morning. And you’ll get a small set of practical levers that reduce uncertainty instead of adding effort, like a simple reachability rule, a boring urgency gate, and a short variance audit you can run in 3 days.

The goal is not perfect bedtime habits. It’s making reachability predictable enough that your nervous system stops acting like it’s on call.

I’m french, born in 1974, and I’ve spent most of my adult life at a desk across Beijing, Berlin, and now Lisbon, often past midnight. When your teams are spread across time zones, “after hours” is a moving target, and randomness creeps in by default. When the system becomes predictable, the brain can finally stand down.

Random reachability keeps the brain on standby

The evening turns into a quiet watch shift

The tell is physical before it’s intellectual: the jaw set, the shoulder tension, the slightly held breath when a notification could happen. You’re not working, but you’re not settling either. Tiny “micro-alarms” keep firing in the background, even with the screen dark.

That monitoring feeling is not just in your head. Research on after-hours availability expectations links extended reachability to poorer detachment and weaker recovery, which often shows up as sleep problems for desk workers (Dettmers, 2017; Sonnentag & Fritz, 2007). Annoyingly, it’s that boring.

A useful way to describe it is micro-alarms. Brief internal status checks. You don’t even have to open Slack. The brain still scans for signals that you should be responsive, like a system waiting for an interrupt.

Once you notice micro-alarms, the next question is what keeps them running. Often it’s not volume. It’s unpredictability.

If messages only arrive in a clear window, the brain can learn to downshift outside that window. If pings are random, the brain stays in listening mode because maybe this time it matters. Intermittent reinforcement helps explain why random cues create stubborn checking habits, even when the “reward” is rare (Ferster & Skinner, 1957).

This is why a lot of generic sleep advice misses the main lever for knowledge workers. It’s not only blue light or “perfect” bedtime habits. It’s uncertainty about whether you might be needed.

Telepressure is the name for that internal push to respond fast even when you’re trying to be offline, and it predicts worse detachment and weaker recovery (Barber & Santuzzi, 2015; Dettmers, 2017). Yes, it’s that small and that dumb.

Evidence you can actually use

Four anchors without turning this into a checklist

A few findings show up again and again.

You don’t need a study dump. The shape of the problem is consistent. Random reachability keeps the nervous system slightly armed.

The standby sleep signature for desk people

How it shows up at night

The triggers can be subtle.

You fall asleep fine, then the night feels like thin sleep. Not dramatic awakenings. Just a sense you never fully powered down. In plain terms it is often a continuity issue, more wake-bouts and time awake after sleep, even if you can’t list them.

It can feel different from anxious overthinking, and that difference matters. The wake-ups are sometimes cue-driven and a bit mechanical.

Research on phantom vibration and ringing shows how expectation can generate false alarms (Rothberg et al., 2010; Drouin et al., 2012).

If you’re the type who checks HRV and sleep scores in the morning, this is where numbers can mislead. Consumer devices often miss brief awakenings and underestimate wake time, so a “good” score can coexist with feeling wrecked. Treat it as a proxy, not a verdict.

The next day feels like personality drift

Mornings can start with a weird reload time. Awake but sticky. The reflex is to scan comms early, not from ambition, but because uncertainty feels physically irritating.

Then attention gets jittery. Work becomes tab-hopping, inbox refreshing, and a sharper tone with low tolerance for “we’ll see.” Fragmentation harms vigilance and increases lapses and variability (Bonnet, 1986; Lim & Dinges, 2010). A jumpy brain likes reassurance.

If that is the output, the input to change is predictability, not motivation.

A 3 day reachability variance audit

Quick test not a new tracking hobby

This is not a full sleep diary or a wearable project. It’s a 3-evening, 3-morning reality check focused on uncertainty. Total time: 30 seconds at night, 30 seconds in the morning.

Evenings log the conditioning signal

Pick a personal cutoff time. Each night, mark 0 or 1.

That middle item matters. Standby-style effects show anticipation can hurt sleep even without a call, and telepressure frames this as preoccupation, not inbox volume (Barber & Santuzzi, 2015).

Mornings log continuity and scanning

Keep it simple. Binary is fine, or 0–2.

This lines up with what fragmented sleep tends to hit first the next day, stable attention and vigilance (Lim & Dinges, 2010).

Interpret variance not totals

Don’t obsess over how many pings happened. Look at variance.

One random late message can teach always listening. The same number of messages inside a known window is easier to contain. Intermittent reinforcement is sticky and hard to extinguish (Ferster & Skinner, 1957).

If randomness is the driver, the best fixes are usually system rules, not prettier bedtime habits.

Remove the randomness

Make evenings legible with 1 routing rule

The brain powers down when the system is predictable. A routing rule makes after-hours contact boring and explicit, which helps detachment in boundary management research (Dettmers, 2017).

The fear is looking unresponsive. The trick is being specific and having an exception path.

Treat it as a reliability feature, not a personality trait. Role ambiguity is a real stressor and research keeps pointing in the same direction (Jackson & Schuler, 1985; Gilboa et al., 2008).

Build a boring urgency gate

Pick 1 channel for true urgency, usually a call or an SMS keyword. Everything else waits for a scheduled reply window.

Then define urgent while you’re tired and your judgment is not at peak performance.

Norms and control can move sleep in real workplaces. In the STAR intervention, shifting work practices and increasing control improved sleep outcomes measured with actigraphy (Olson et al., 2015). Translation for desk work: expectations and control are not decoration.

Cue control without screen rules cosplay

Move the portal.

If the phone is within reach, it becomes the fastest uncertainty reducer. Even the mere presence of a smartphone can reduce available cognitive capacity, so distance is not just symbolism (Ward et al., 2017). A concrete version is charging it across the room instead of on the bedside table.

Next, make notification rules unambiguous so you don’t live in “maybe.” Partial filtering can be the worst of both worlds because the brain stays in signal-detection mode. It is often cleaner to go fully off, or allow only the 1 deliberate emergency channel.

Close the loop with a deliberate last check time. The aim is not purity. It is closing the uncertainty tab so you don’t renegotiate with yourself every 12 minutes. Batching checks can reduce stress in field research (Kushlev & Dunn, 2015).

When you can’t disconnect

A late window that makes checking boring

A workable compromise is a single late check window. 10 to 15 minutes around 21:30, then off.

This trains timing, not willpower. If checking only happens in that small slot, the brain stops sampling the channel every time unease spikes. Rate limiting interrupts, but for your sleep.

And if it is truly an on-call week, formalize it as much as you can so non on-call nights are actually off. Standby research is pretty clear that anticipation alone can fragment sleep. Half on call every night is the worst mode.

A status note that prevents the evening pings

Another lever is reducing inbound messages by removing other people’s uncertainty. A short status note before dinner can prevent the quick check “are you around” pings. Clarity reduces role ambiguity and can reduce telepressure by making expectations explicit (Jackson & Schuler, 1985; Gilboa et al., 2008; Barber & Santuzzi, 2015).

A template keeps it fast.

Status: X is done, Y is next. Next update at 09:00.
If blocked now call or SMS. Otherwise I’ll reply in the morning window.

State, next time, escalation path. Calm tone, no apology, no extra availability promises.

What improvement looks like when life stays messy

Useful wins are boring.

If anything gets measured, measure trends with guardrails, not nightly grades. A simple “1 block” rating is often enough.

Async tools plus ambiguous urgency is a perfect recipe for telepressure. Telepressure is expectations turned into internal monitoring (Barber & Santuzzi, 2015). Changing the system people operate inside can improve sleep outcomes, not through heroics, but through clearer norms and control (Olson et al., 2015).

Across Beijing, Berlin, and Lisbon, the same pattern keeps coming back. Predictability lets the brain stand down. Make reachability boring and explicit, and sleep gets a fairer chance.