III. Ordinary Things Essay 20
The Phone in Your Pocket
Attention as the thing being traded
Your phone is in your pocket. It pings. You take it out, you look at it, and the thing on the screen is a fact you already knew, or a picture of somebody's breakfast, or a sale on something you were not going to buy. The whole interaction took about eleven seconds. It also told the app that you looked, which is the part that matters to the app and the part you did not think about at all.
Here is the thing to understand, and it is a mechanical thing rather than a moral one. Every design decision in that phone, taken together, is aimed at one quantity. Not at your happiness, and not at the quality of the service, and not even at your revenue. At the number of seconds your eyes spend on the screen, because that number is the raw material of the business.
Attention is the thing being traded, and the system is shaped to get as much of it as it can. Once you can see the shape, the individual decisions stop looking like a series of separate choices and start looking like one thing, which makes them much easier to argue with.
Take one feature. A messaging app sends a notification for every message, including the ones from a group of six people you have muted, and including the ones that say "haha" or "ok". Each of those notifications is a small decision, made for a good reason: it makes the app more useful to somebody, often the person receiving the message. The engineering reason is sound: the notification is fast, cheap, and it keeps people in the app.
Now look at the loop rather than the feature. The notification arrives, you look, the app receives a signal, and the signal makes the app more likely to send you something. The signal is not about the message. It is about you. Over weeks, the app learns that a message from that group produces a lookup, so it sends those messages with more urgency, which produces more lookups.
Nothing in this description says the app is manipulative. Every individual feature is defensible. The picture only appears when you draw the loop, and the loop has a direction built into it, and the direction is toward more of the thing that is being measured. The app is not trying to be addictive. It is trying to maximise a proxy, and the proxy is a decent one, and a decent proxy maximised without limit goes to the ceiling, and the ceiling is a person looking at a screen in a dark room at one in the morning.
The trade
Every app on your phone is running the same kind of bargain, and the bargain is easier to see once you name both sides of it. On one side, you get something real: a message from a friend, a bank balance, a map, a photograph, a game, a bus time. On the other side, the system gets your attention, and your attention is what it converts into whatever it is actually for: advertising income, subscription revenue, purchase data, and in some cases the engagement number itself, which is the thing that makes the advertising worth buying.
The word that gets used here is engagement, and it is worth noticing how carefully it is chosen. It does not mean satisfaction. It does not mean the user got what they wanted. It means the user kept going. A metric that measures whether the user came back is measuring something, and that something is not the same as whether they were glad to.
This is the same structure as the mark on the exam paper, and the same structure as the number in a report card, and it should be recognised as one. A proxy stands in for something real, the proxy is easier to count, and once the proxy is being counted, the thing it stands for starts to drift. A bus company measures passenger numbers, so it puts on more buses and they get fuller, and nobody needs to do anything wrong for the service to get worse. A school measures exam results, so it teaches to the exam. An app measures minutes, so it makes minutes.
There is one more thing on the right-hand side of the bargain that is usually left out, and it is not your data or your money. It is your attention taken away from everything else. The notification does not only cost you eleven seconds. It costs you the eleven seconds, plus the getting back into whatever you were doing, which is not eleven seconds and is not a small cost. The interruption is the product. The thing on the screen is nearly irrelevant, and the fact that people sometimes forget what was on the screen is not a bug in the design.
Why the loop runs hot
The loop in the notification example is a reinforcing loop, and reinforcing loops have two properties that make them hard to argue with. They compound, and they are invisible from the inside.
The compounding is easy to see in hindsight and nearly impossible to see at the time. A feature that adds a minute a day adds six hours a month. The change from the app you had in January to the app you have now is not the sum of its features. It is the sum of its loops, running at once, for months, in a direction none of them was designed to push individually.
The invisibility is the other half, and it is the reason an individual person is not the right place to look for the cause. You are not being manipulated. You are not weak. You are a person with a functioning reward system responding to an engineered signal, and the engineering is not a conspiracy. It is a set of very reasonable decisions, made by very competent people, each of which increased a number. The number is minutes. The minutes are what the business is made of. It is not that somebody decided to make you check your phone more. It is that everybody decided to make the number go up, and the number going up is what they were paid for.
What if the app simply did not notify you? A message arrives, no banner, no sound, no badge. You find out when you open the app. Everything still works. Every message still arrives, every call still comes, every update is still there. What changes is that the app no longer gets to interrupt you, and the interruption was the thing it was actually buying.
Now change the condition. What if the notifications were batched, so that everything that arrived during the day was waiting when you next opened the app, with the time it arrived on it? Almost all of the value of the notification is the timing. The message is the same message. What the notification sold was the removal of the gap between the message arriving and you finding out, and if you are going to look at the app anyway, that gap is most of the value, and the rest of the value is the interruption.
"Technology is neutral, it is the people who misuse it." The tools are not neutral, and the reason is not moral, it is mechanical. A tool that is faster at one thing makes that thing more common, and the commonness changes the system the tool sits in. A hammer does not want you to hit nails. A thing that makes interruption cheap produces more interruption, and it does so whether or not anybody intends it.
The other misconception: "you just have to be more disciplined." Discipline is a real thing and it helps, and it is also the least interesting answer, because it puts the whole burden on the person with the least power in the arrangement. If avoiding eleven-second interruptions required constant vigilance, the arrangement would be working exactly as designed. The design question is not about your character. It is about how many decisions a week it takes to keep a device in a particular state, and whether that number is one or forty.
What the system is for
Going back to the level above the ladder, there is a purpose question hiding in all of this, and it is the question that actually decides what happens.
What is a social feed for? Not for connection, whatever the people who build them say, and certainly not for the users, because the users did not choose it and cannot leave it without losing the people on the other end. A feed is for engagement, because engagement is the product. The stated purpose and the actual purpose differ, and the actual one is the one that has been optimised for years, and the design of the thing follows it. Everything about the interface, from the pull to refresh to the streak counter to the unread badge, is an answer to "what makes you come back", and the answers are consistent, which is how you can tell that there really is a purpose underneath.
And once you have said that, the question of what you want is no longer a question of willpower. It is a question of arrangement. Which apps are on the phone. Which ones notify. Which ones are on the home screen rather than in a folder. Whether the phone is in the room when you are working, or in the other room, or face down and out of reach. Each of these is a decision about the structure, and each of them is much easier to make once and harder to undo later.
What to do with this
Three things, in the order that works.
First, find the loop rather than the app. Do not ask which apps are a problem. Ask what the loop is for each one, and which of your loops is currently running. Almost everybody has one app whose loop they have never looked at, and it is almost never the one they would have guessed, because the one they would have guessed is the one they have already thought about.
Second, change the setting rather than the intention. A promise to yourself not to check the phone is a resolution, and resolutions are answered by the same part of you that makes them. A setting is a decision about the arrangement, and arrangements do not require willpower to maintain. This is the whole argument of the level above the ladder, applied to a small and rather embarrassing subject.
Third, put the phone somewhere else. Not face down. Somewhere else. A room with no phone in it is a different system from a room with a phone in a pocket, and the difference is not in you, it is in how many times the loop gets a chance to run.
Limits
This essay deliberately says nothing about your health, your sleep, your attention span, your dopamine, or your mental wellbeing, and it makes no claim about how much time people spend on their phones. Those are questions with real research behind them and this site does not have the standing to answer them. What is described here is the mechanism, and the mechanism can be read from the design without any claim about what it does to a person.
Second limit: the picture is a simplified one. Not every design decision in every app is aimed at the same number. Some are aimed at reliability, some at safety, some at battery life, and some at a different number entirely that a different group of people is measured on. The claim is not that the designers are villains or that the system is secretly sinister. The claim is that a system with a purpose produces a consistent shape, and the shape is visible.
Third limit: the loops described here are the ones that work on almost everybody, and they are not the only ones. The advice at the end works for the common case. For somebody whose work is to make a phone app work, the loop is the point, and the answer is not to switch off the notifications but to build a better one, and this essay is not for them.
Stuck
You are stuck at the system level with a phone when you can describe the loops perfectly and the phone is still on the table at every meal. Understanding the mechanism has not changed the arrangement, and the arrangement is what produces the behaviour.
The way out is to change the arrangement, and to do it in the smallest way that will hold. Not "I will use my phone less", which is a resolution about a quantity, but "the phone sleeps in the hall from eight", which is a decision about a place. The second one can be forgotten, because it does not need remembering, and the first one will be forgotten by eight thirty, because it is a promise and promises are answered by the part of you that is already tired.
Sources
- No specific study, statistic, company, or product is named or cited in this essay, and no claim is made about time spent, health, sleep, or wellbeing.
- The description of the attention loop is the author's own, and is offered as a way of reading the design of a familiar class of software.
- The examples are ordinary and hypothetical. Where a notification example is used, it is a composite rather than a report of a particular app's behaviour.