On Being Reachable About Everything

The sound arrives before the meaning does. My phone buzzes, or one of the small pager apps chimes, and for about half a second I have no idea whether it's an alert from the training rig, a message from a friend, or something that needs nothing from me at all. The body reacts anyway. Shoulders tighten. Attention snaps sideways. Whatever I was doing loses a small tax to the interruption before I've even looked at the screen. Then I look. It's usually nothing. That gap between the sound and the information is where most of the cost sits, and it's the same gap whether the notification turns out to matter or not.

I've written before about checking long training runs compulsively, refreshing a log window that had nothing new to show, as if watching harder would move the numbers along. I've also written about building better alerting into NOGnet and Magatama specifically so I could stop doing that, with thresholds and summaries that only page me when something needs a decision from me. That part worked well. What it didn't touch was the flinch. The alerts got quieter. My reaction to the sound of a notification, any notification, stayed about where it was.

There's a difference between being reachable if something is genuinely wrong and being reachable about everything, all the time, by default. The first is a safety net. The second is a low hum that never fully switches off, and it's the one I'd built for myself without quite noticing, one integration at a time. Every system I run pages me. Every chat has push notifications turned on. Ordinary life adds its own layer on top of that: a delivery update, a message that could have waited a day and didn't. None of it is urgent by itself. Together it adds up to a permanent low-grade readiness that never resolves into either off duty or on call, just an undifferentiated maybe.

What I've changed is smaller than I'd like it to be. Quiet hours on my phone now block almost everything overnight except a short allowlist for the handful of things that would really justify waking up. I trust the automated triage on the training rig enough that I no longer open the dashboard just to look, most days. Both of those took real effort to set up and then more effort to leave alone once they were running, because not checking felt like negligence for a while before it started feeling like trust.

What I haven't fixed

The reflex is still there. A notification sound gets a physical response before my brain has any content to judge it against, and no amount of good filtering upstream changes that first half-second. I've also noticed I still treat “I could check” as functionally the same as “I should check,” which quietly defeats a lot of the point of building triage in the first place. Having a system I trust doesn't automatically make me trust it in the moment. It gives me the option to, which counts for something, without making the anxiety go away on its own.

I don't think the answer is fewer systems. I like running them, and most of the alerts are ones I designed on purpose, for reasons that still make sense. What I keep circling back to is the assumption sitting underneath all of it, the one that treats availability as a virtue with no ceiling, where more reachable is always better than less. Quiet hours are a small, mechanical rejection of that idea. They don't feel like much at the moment I set them. They feel like something the next morning, when I wake up and nothing broke while I wasn't looking, and I realize I didn't need to know that in real time for it to be true.