Dipperstein All articles
Deep Dives

We Log Every Request Except Our Own Lives: Notes on Personal Telemetry

Priya Raghavan
We Log Every Request Except Our Own Lives: Notes on Personal Telemetry

Photo: vintage brass bucket and well rope at dusk, moody lighting, via stylebaggage.com

Here's an uncomfortable exercise: name the last system you instrumented. For most of us it's easy. We've got dashboards for p99 latency, alerts on error budgets, traces stitched together across a dozen services. I once spent two full days tuning a histogram bucket for a service that handles, generously, forty requests a minute.

Now the harder question: name the last metric you collected about yourself. Not a step count your phone guilt-tripped you with. An actual measurement you chose, defined, and acted on.

For me the answer was "nothing," and it stayed nothing for about fifteen years of professional work. That gap is what I've come to think of as the diving bell reflex — we build elaborate apparatus to sample everything in the world around us, then never once dip the bucket back into our own well.

The Blind Spot We Keep Shipping

Engineers are professional observers. We add logging like other people add salt. And yet the decisions that shape our careers most — what to learn next, which project to take, whether this job is still good for us — get made on vibes, anecdote, and whatever happened this week.

The asymmetry is almost funny. At work I know the uptime of a cron job to four nines. I could not tell you, before last year, how many side projects I'd abandoned per year, or how my time actually split between writing code and sitting in meetings. I had opinions about all of it. Opinions, it turns out, are what people have when they don't have data.

There's also a failure mode in the other direction worth naming: the quantified-self crowd that tracks 47 metrics and drowns in them. That's not telemetry, that's hoarding. The point of a good production dashboard isn't that it measures everything — it's that it measures the few things someone would actually do something about. Personal telemetry deserves the same discipline.

The Setup: Four Counters and a Calendar Block

So last summer I ran an experiment on myself. The rules were deliberately boring:

  1. Track no more than five counters. If I wanted to add one, I had to retire another.
  2. Manual entry beats automation. This sounds backwards, but a minute of manually tallying forces you to actually see the number.
  3. Thirty minutes, every Friday, to review. Calendar block, recurring, non-negotiable. The review is the product; the counters are just inputs.
  4. No streaks, no gamification, no apps. A plain markdown file. Apps optimize for engagement, and engagement is not insight.

My five counters, for the record: hours of deep work (honest estimate, not fantasy), number of things I said yes to that I wished I'd declined, side-project commits, hours of reading that wasn't documentation, and a single subjective "drain score" from 1–5 at end of week.

Total overhead: about four minutes a day. That's it. That's the whole practice. If a personal telemetry system needs more scaffolding than a sticky note and a Friday alarm, it will die in month two, guaranteed.

Six Months In: What the Numbers Actually Said

I expected the data to confirm what I already believed. It mostly didn't.

Finding one: my sense of "busy" was a latency metric, not a throughput metric. I felt busiest in weeks with the most interruptions, but the counters showed my deep work hours were lowest exactly when I felt most slammed. Feeling busy and being productive were anti-correlated, and no amount of introspection would've surfaced that. Once I saw it, I started defending mornings like they were an on-call rotation — because functionally, they are.

Finding two: yes-fatigue was measurable. The "wished I'd declined" counter averaged about one per week. That doesn't sound like much until you multiply it by a year and realize that's fifty commitments my past self signed my future self up for. Seeing the tally made saying no feel less like rudeness and more like capacity planning.

Finding three: the drain score predicted decisions better than anything else. Two consecutive weeks of 4s meant I was about to make a bad call — usually an impulsive "maybe I should interview somewhere" spiral or an overengineered rewrite of something fine. The number didn't fix anything. It just gave me a check-engine light, and check-engine lights are genuinely useful even when they can't tell you what's wrong.

Finding four: what didn't move. Side-project commits barely budged all six months. I'd half-hoped tracking would shame me into building more. It didn't, and honestly that's useful information too — it told me the bottleneck was desire, not visibility.

Sampling Yourself Without Becoming a Lab Rat

I want to be careful here, because this practice can curdle fast. Measure yourself like you'd measure a system you're stewards of, not one you're optimizing for someone else's SLO. Three guardrails kept it healthy for me:

The Well Doesn't Run Dry, But Somebody Has to Dip

There's a reason this site is called what it is. The whole job, whether it's a data lake or a production cluster, comes down to the same gesture: lower the dipper, pull up an honest sample, look at what's actually in the water. We do this reflexively for every system we touch at work.

The one system we're guaranteed to have for our entire career sits right on the other side of the keyboard. Four minutes a day and one calendar block. That's the entire barrier to entry. The well's been there the whole time — somebody just has to drop the bucket.

All Articles

Related Articles

The Kettle, the Ladle, and the Compiler: A Soaking-Hot Tour of Type Strictness

The Kettle, the Ladle, and the Compiler: A Soaking-Hot Tour of Type Strictness

My Well Pump Is the Worst Database I've Ever Administered

My Well Pump Is the Worst Database I've Ever Administered

The Big Dipper Doesn't Move, But My Tripod Sure Did: A Star Tracker Built Out of Junk Drawer Physics

The Big Dipper Doesn't Move, But My Tripod Sure Did: A Star Tracker Built Out of Junk Drawer Physics