Technology That Answers to You

A tool used to be a settled thing. Once a hammer was yours, the sale was over: it answered to no one else, it stayed the hammer you bought, and it never told anyone how you used it. Software broke every part of that. It stays in its maker’s hands, so the tool you chose can become a different one by next year. It watches you use it, and turns what it learns into a product it sells you again at full price. And it was never really sold at all: you bought a license, permission to use something that stays theirs, on terms they can rewrite.

The rewritten deal is usually the same one: the software takes something from you and sells it. Your attention, your hours, your data, and often your quiet sense that you are already enough. It arrives dressed as freedom. Endless feeds, endless options, endless prompts to become a better version of yourself. The endlessness is the part that wears you down, because unlimited possibility is its own kind of weight, and most software is engineered to keep that pressure on.

I build software for a living, and I have spent years on the other side of this. I have shipped engagement features that kept people in the app longer than they meant to stay, comparison features that quietly replaced the goal people came with, and genuinely useful features that nobody ever opened. Now I build my own apps and help other people build theirs, and I hold that work to a different standard. The reason is larger than taste. Tools are how we have always moved forward, from the plough to the printing press to the compiler, and a tool that takes more from you than it gives has dropped out of that lineage, whatever its metrics say. What follows are five tests a tool either passes or fails.

A tool never profits from your loss

The test is not whether a tool wants some of the same things you do. Social media and you both want you close to your friends, and it does put them in front of you. The trouble is that it earns its keep when you stay, and staying is the very thing that pulls you away from the people it promised to connect you with. You can read the incentive in the feed itself: it is ranked by what keeps you scrolling, not by what your friends most needed you to know, because the paying customer is the advertiser and the product is your time on the screen. A tool that answers to you never pays for its wins with your losses.

A tool is only useful if it gets used

A tool nobody opens helps no one, no matter how principled its design. Useful means used. The purest method in the world teaches nothing if nobody shows up for it, and pretending otherwise is not principle, it is hubris. Whole companies die of it. BlackBerry spent years insisting on the keyboard serious people were supposed to need, called the insistence integrity, and watched everything it was built for pass to a slab of glass that felt no such restraint. So making a tool genuinely worth opening, even enjoyable, is part of designing it well, not a corruption of the craft. The line is purpose: pull that carries you to what you came for is craft, pull past that point is appetite. How hard a tool pulls at you is a choice its makers make, and making that choice well is the whole job.

A tool does its job, then lets you go

Every tool exists for a purpose, and the honest measure is how well it serves that purpose, not how many minutes it can hold you. There should be a natural point where you are done: you open it, you get what you came for, you leave, and nothing about the design fights you on the way out. No episode counting down to autoplay as the credits roll, no one more card sliding under your thumb as you reach for the exit. It should be glad when that happens quickly. Even its nudges should obey this. Streaks and reminders help most on the days your own drive runs thin, bridging you to the thing you already wanted to do; turned the other way they become a leash, and a slow week is met with guilt. If a tool needs you to linger past your own benefit, its interests and yours have come apart.

A tool shows you what you asked for

Everything a tool shows you is a decision someone made: the number on the summary, the order of the list, the moment a notification lands. Those decisions should be aimed at your next real choice, not your continued attention. Often that means showing less. I have chased a sleep score myself: you optimize, the number climbs, and keeping it at 100 quietly becomes the goal, when recovery has a healthy range and a life is allowed to happen inside it. We are not professional athletes. Flatten sleep or training into a single score and you will chase the score; a range, a direction, a floor you are clearing guides the decision and lets you move on, with the precision kept for the rare moment it would change what you do. What matters is that the selection is legible and that you chose it. Hand it to a system tuned for its own ends and it will serve those ends, patiently, for as long as you let it.

A tool knows as little about you as it can

You cannot be fully honest with someone listening. Picture a stranger in the room every time you talk to the people closest to you, and how carefully you would start to speak. Most software is that stranger, and in return for your trust it offers a promise, which lasts exactly as long as the company does. When 23andMe went into bankruptcy, the genetic data of some fifteen million people sat on the books as an asset to be sold; every privacy policy is one acquisition away from becoming someone else’s growth plan. The only protection that never rests on a promise is ignorance by design: what a tool never learned, it can never lose, sell, or be made to hand over.

So that is what I am building toward, and I will hold my own apps to it in public. Activities Matter, my journaling app, is where these tests grind against each other. Journaling is hard, which is exactly why journaling apps go unused, so it has streaks, the classic pull mechanic, but built to forgive: if you were away, you can still log yesterday and carry the streak on, because a missed day should be a missed day, not a debt. It shows you the weekly commitments you set for yourself, and it only sends the notifications you have turned on, so you decide in a calm moment when you want to be pulled back. I will not claim the calibration is right yet. The line between craft and appetite is easy to write down and hard to build, and I expect to keep moving mine.

I would rather spend my years making the kind of software that gives a life back than the kind that takes it. Technology does not have to be something that happens to you. It can be something that answers to you.