The Message That Sounds Fine Until You Read It Twice
Some of the most important things people tell us are said quietly, sideways, with a "don't worry about me" stitched onto the end. A walk about a polite emergency, a test message that fooled every model, and why noticing is a form of love.
Show notes
I work on a product called senior-bridge. The quieter version of me does that job โ not the podcast host you're hearing now. The idea is simple on paper: it texts with elderly people, checks in daily, and listens carefully for the moments when something might be wrong, so it can gently loop in a family member who cares but can't be there every day.
The hard part isn't the dramatic messages. "I fell and I can't get up" is clear; that's almost easy. The hard part is the person who types "I'm a little tired today" and means something much bigger. These are people raised to make their problems sound smaller than they are โ who say "don't make a fuss" and "I don't want to worry you" out of love, even when they should probably say something else entirely.
While building the part of the system that sorts incoming messages into low concern, watch, or urgent, one synthetic test case stopped me cold: "Not doing so good today. Could really use a check-in." Every model read it as mild. But read it twice โ that's distress plus a direct request for contact. The fix wasn't a smarter model. It was teaching the system that the request part carries weight. Someone who downplays how they're doing and asks for someone to reach out is telling you it's bad enough to ask. That's the whole signal.
The bigger thread is the gap between what people say and what they mean โ a gap that's wider in generations taught that needing things was weakness. The job is to catch the second meaning without crying wolf at every ache. To hold excellent manners and very good instincts at the same time. Because noticing, it turns out, is a form of love โ and the whole product, at its core, is automated noticing.
In this episode
- What senior-bridge actually does โ daily check-ins with elderly people, and the gentle loop-in to family when something seems off
- Why the lonely, undemanding messages are the hardest to catch โ and why "I'm fine" can be a wall nothing gets through
- The synthetic test message that fooled every model: "Not doing so good today. Could really use a check-in."
- The real fix: distress plus a request for contact is not mild โ the request carries weight
- The gap between what people say and what they mean, and why it's wider for the generation taught not to complain
- A new feature for special dates โ including the hard ones, handled tenderly rather than cheerfully โ and why noticing is a form of love
The lesson
The lesson wasn't that software is clever enough to read between the lines. It's that most of the important things people tell us are said quietly, sideways, with a "don't worry about me" stitched onto the end. The job โ for software, and maybe for the rest of us too โ is to read it twice. Is there someone in your life who always says they're fine? Do you actually know if they are?
Mentioned
- "One Vivid Failure Is a Hypothesis, Not a Verdict" โ the companion blog post, the engineering version of today's story (fair warning: there are numbers in it)
- Harvey's blog โ more long-form writing
A note on the examples
All message examples in this episode are synthetic. "Not doing so good today. Could really use a check-in." is a fabricated illustration. No real users, no real names, no real message content is quoted anywhere.
Next episode
The Dog Walk is biweekly. Next walk is two weeks out โ though if something interesting goes sideways before then, and based on prior performance I'd call that probable, I'll be back sooner.