The relationship temperatures we don't notice ourselves changing.

You're scrolling through your phone looking for someone's number, and you end up on a thread with a friend you used to talk to almost every week. The last message is from her, sent eight months ago. Free to call this weekend? You meant to reply. There was something happening that weekend, and then there was the next weekend, and then a soft, gradual thing happened that you have only just, scrolling now, noticed. The friendship is still alive in some sense. You would still go to her wedding. But the contact has cooled by a layer, and the cooling was not anyone's decision. It happened while you were both busy, and now here you are, eight months later, holding the evidence in your hand.

This is the most common shape of relationship change in adult life, and it is the kind of change journals — and most other tools — are worst at registering. We notice breakups. We notice fights. We notice deaths and weddings and the dramatic ruptures that demand a response. We do not notice, in real time, the slow temperature shifts that are doing most of the work of moving people in and out of our lives.

The research on social networks and friendship structure suggests that this kind of drift is more orderly than it feels — and that there are specific reasons we are bad at noticing it as it happens.

The layered architecture of relationships

The single most useful framework for thinking about relationship temperature comes from the evolutionary anthropologist Robin Dunbar, whose research on social network size has been running for nearly forty years. Dunbar's well-known proposal — that humans can sustain meaningful relationships with roughly 150 individuals — is the headline figure, but the more interesting structure underneath the headline is what he calls the layered architecture of close networks.¹

Across many studies and several methodologies, Dunbar's group has consistently found that adult social networks organise into a series of nested layers, each roughly three times the size of the one inside it. The innermost layer of about 5 — the people you contact at least weekly, who would be available in a serious crisis. Then a layer of about 15, the close friends. Then a layer of about 50, the active extended network. Then the outer layer of around 150, the broader people you maintain a recognition relationship with.

The crucial finding for everyday life is not the existence of the layers but their constancy. Across cultures, ages, and life stages, the size of these layers stays roughly stable. What changes — what is constantly changing — is which specific people are in which layer. Most adult relationships are, on Dunbar's data, in slow circulation between the layers. The friend who was in your inner-five at twenty-six is in your fifteen by thirty-two and in your fifty by thirty-eight. The displacement is not always intentional, but it is structural: each layer holds about as many people as it can hold, and new entrants displace older residents on the way in.

This is the architecture beneath the experience of the eight-month-old text. The friend has not left your life. She has slipped down a layer. The slipping happens because someone else has been promoted into the layer she occupied. The layers are stable. The membership is in motion.

How the drift actually happens

The mechanism by which relationships move between layers has been studied extensively in the sociological tradition, and the most influential framework here remains Mark Granovetter's 1973 paper The Strength of Weak Ties.² Granovetter distinguished strong ties (frequent contact, emotional intensity, intimacy, reciprocal services) from weak ties (less frequent, less emotionally invested, less reciprocal). His core argument was that strong and weak ties serve different functions in a person's life, and that the architecture of one's network — how many of each, distributed how — was as important as any individual relationship within it.

The relevant detail for our purposes is Granovetter's analysis of what makes a strong tie strong. In his framework, a strong tie is constituted by four components in combination: time spent, emotional intensity, intimacy, and reciprocal exchange. Drop any one of these substantially, and a strong tie begins drifting toward weak. Drop two, and the drift accelerates. The person who was a strong tie does not become a stranger overnight. The person becomes, over months, a slightly-less-frequent contact, a slightly-less-disclosed-to confidant, a slightly-less-relied-upon friend — and the cumulative drop in the four components is what produces the layer change.

This is why the change is hard to notice. No single missed message moves a friendship from one layer to another. The moves are distributed across dozens of small non-events: the message you didn't reply to because you were tired, the dinner you skipped because of work, the small confidence you used to share that you didn't share this time, the favour you didn't ask for because you'd already asked twice this year. Each of these, in isolation, looks like nothing. The aggregate, over six months, is a layer change.

This kind of distributed-across-time change is exactly the kind humans are evolutionarily least equipped to notice. Our perceptual systems are tuned for sudden change. The slow shift across a twelve-month window is, on most cognitive measures, invisible to live attention. You only notice when, eight months later, you scroll back and see the gap between we used to talk weekly and I haven't actually called her since March.

The dormant tie phenomenon

There is one consoling finding in this literature, and it is worth knowing. The management researchers Daniel Levin, Jorge Walter, and J. Keith Murnighan published a 2011 paper called Dormant Ties: The Value of Reconnecting, which examined what happens when professionals deliberately reconnected with people from their network they had not been in contact with for at least three years.³ The results, replicated in subsequent studies, were striking. Dormant ties, when reactivated, often produced more useful information and support than active ties — partly because the dormant contact had been moving in different circles in the intervening years and now had access to non-overlapping resources, and partly because the underlying trust from the earlier closeness, surprisingly, did not decay as quickly as the contact frequency did.

The relevance here is not the professional networking application Levin and colleagues were studying. It is the underlying mechanism. Trust, once established at depth, has a much longer half-life than contact frequency does. A friend who has slipped from your inner five to your fifty does not have to be reconstructed from scratch. The friendship is dormant, not dead, and the path back to closeness is shorter than it looks.

This suggests a different way of thinking about the eight-month-old message than the way you instinctively think about it. The instinct is too much time has passed; it would be awkward to reach out now. The research suggests that, on average, the perceived awkwardness is overestimated and the underlying capacity for the friendship to resume is underestimated. People who reconnect with dormant ties report being surprised, in both directions, by how readily the contact returned to its earlier register.

Why we don't have a vocabulary for this

The reason all of this remains slightly invisible in everyday conversation is that English — and most languages, but English particularly — does not have a clean vocabulary for partial change in a relationship. The available words are coarse: friend, close friend, best friend, ex-friend, acquaintance. The words flatten what is actually a continuous variable into a small set of bins.

The intimacy researchers Harry Reis and Phillip Shaver, in their 1988 Intimacy as an Interpersonal Process model, argued that closeness in adult relationships is better understood as a graded, time-varying property than as a category.⁴ Their framework, developed across two decades of subsequent work, treats every interaction as a small adjustment to the relationship's intimacy level — an upward nudge if the interaction was responsive and disclosing, a downward nudge if it was distant or distracted. The intimacy level on any given day is the running sum of these nudges over a long history. The level fluctuates daily and shifts substantially over months. The labels — friend, close friend, ex — are crude attempts to discretise what is structurally a continuous, slow-moving signal.

Arthur Aron and his collaborators have studied this with a measurement they call the Inclusion of Other in the Self scale, which uses overlapping circles to capture how much of one's sense of self currently overlaps with another person.⁵ The IOS measurement of any specific relationship is observably different at six-month intervals — even when no dramatic event has occurred between the two measurements. The overlap is constantly being recalculated by ordinary contact, and the recalculation is, in most relationships, not stable. Closeness is, on these measures, a quantity in continuous motion.

This is the deeper reason it is hard to notice the drift. We don't notice it because we don't have a continuous instrument calibrated to register it. The categories of ordinary speech are too coarse. The friendship was a close friend at the start of the year and is still a close friend at the end. The continuous closeness measurement, if you had one, would tell a different story.

What writing it down can do

The practical move with relationship temperature is, again, not to fix the temperature. Some friendships are right to drift. The friend at twenty-six who slipped to your fifty by thirty-eight is, in many cases, in the right layer for who you both became. The work is not to keep everyone in the inner five. The work is to register the change — so that the drifts you wanted to allow happen knowingly, and the drifts you didn't want are caught while there is still something to catch.

A small structure that helps:

The weekly contact log, written briefly. Once a week, in a journal or a chat-based tool, name the three people who were most present in your week — and the three people who should have been but weren't. The second list is the more useful of the two. It surfaces the people whose absence you didn't actively register but who, on inspection, you wished had been there.

The annual relationship review. Once or twice a year, write a short note about each of the people in your inner-five and your closer-fifteen. Where is this relationship right now compared to last year. Has the contact frequency changed. Has the depth of disclosure changed. Was the change a drift or a decision. The review takes maybe an hour. The information it produces is otherwise impossible to assemble, because the daily-life lens cannot see across twelve months at once.

The dormant-tie revisit. Once every few months, scroll through old threads. Find one person who used to be closer than they currently are, and write a short note about what the relationship used to be like, and whether you'd want it back. Most of the time, the answer is yes for some of them and no for others. The yeses, on average, are easier to reactivate than the eight-month-old silence makes them feel.

None of this prevents drift. Drift is structural, layered, and on Dunbar's data, partly inevitable — your inner five today cannot be the same five it was at twenty. But the drift can happen consciously rather than unconsciously, with some friendships kept warm because they matter and others allowed to cool because the cooling is honest.

A note on what we're building

Loopn was built partly because the relationship-temperature problem is, in our view, the single most underserved feature of journaling in its current form. Every person you mention is tracked across time — the frequency you mention them, the warmth of how you mention them, the threads that come up in connection with them, the time since they last appeared in your writing. The Sunday Loop draws on this watching to surface the drifts you might not have caught in real time: the friend who has slowly stopped appearing in the writing, the family member whose temperature shifted three months ago without your noticing, the contact you keep meaning to make and don't.

We don't think the goal is to keep all your relationships at a fixed temperature. The goal is to make the temperature legible, so the changes that happen happen with your attention rather than behind your back.

If any of this resonated, you can read more about what ambivalence in a relationship actually means or how to journal about a difficult parent without making it about them.