The Liking Gap: Why the Small Talk You Dread Is Doing Real Work

There is a specific micro-calculation most of us run several times a week. The lift doors open, a neighbour is already inside, and in the half-second before they close you decide whether to say something about the weather or to look at your phone. The phone almost always wins. Not because you dislike the neighbour. Because a small, confident voice says: this will be awkward, they don't want it either, and you will do it badly.

That voice is wrong about all three things. Which would be a footnote, except that the same voice runs at the school gate, in the office kitchen, at the end of a work call when there are ninety seconds left, and after a while the accumulated small avoidances add up to something that looks a lot like isolation but never felt like a decision.

I say this as someone who has spent most of his adult life on the introverted end of the scale, in a profession full of people who chose it partly because it lets you communicate in writing. So this is not a piece about becoming a person who loves chatting. It is about a measurement error — a real, replicated one — and what you might do with it.

The liking gap

In 2018, Erica Boothby, Gus Cooney, Gillian Sandstrom and Margaret Clark published a study in Psychological Science with a plain enough title: The Liking Gap in Conversations: Do People Like Us More Than We Think?

The design was simple. Strangers were paired, they talked, and afterwards each rated how much they had liked the other person and how much they thought the other person had liked them. The second number came in consistently below the first. People systematically underestimated how positively their conversation partner viewed them, and how much the partner had enjoyed the exchange.

What makes it hold up is that the gap survived the obvious objections. It appeared in brief laboratory conversations and in longer ones. It appeared among university students getting to know new dormitory neighbours, across months rather than minutes. It appeared in workshops among adults who had every reason to be socially calibrated. The people who were most liked were often the most convinced they had underperformed.

The mechanism the authors point to is an asymmetry in what you have access to. Your own inner commentary during a conversation is loud, continuous and unflattering — that came out wrong, why did I mention the parking, they checked the clock. The other person's inner commentary is invisible to you. So you compare your unedited internal draft against their calm exterior and conclude, reasonably enough, that you are the one struggling. They are running the same comparison in the opposite direction and reaching the same conclusion.

Two strangers walk away from a decent five-minute conversation each privately convinced they were the weaker half of it.

The forecast is wrong before the conversation, too

The liking gap describes what happens after. There is a matching error before.

Nicholas Epley and Juliana Schroeder ran a set of studies on Chicago commuters, published in the Journal of Experimental Psychology: General in 2014 under the title Mistakenly Seeking Solitude. Commuters on trains and buses were assigned either to strike up a conversation with a stranger, to sit in solitude, or to commute as usual. A separate group was asked to predict how each condition would feel.

The predictions were confident and backwards. People expected the conversation condition to be the least pleasant. It turned out to be the most. And a detail I find more telling than the headline: participants also assumed that strangers would not want to talk to them, which is why they expected the attempt to fail. In practice, essentially nobody was rebuffed.

Gillian Sandstrom and Elizabeth Dunn found a related effect in a 2014 paper on weak ties, including a study in which people were assigned either to have a genuine brief exchange with a coffee-shop barista or to be as efficient as possible. The ones who had the small human interaction reported a better mood and a greater sense of belonging afterwards. Thirty seconds of ordinary politeness, measurably doing something.

Put these together and the shape of the error is clear. We are bad at predicting that we will enjoy it, bad at predicting that the other person wants it, and bad at judging how it went once it is over. Three biases pointing the same direction, all of them arguing for the phone.

The beautiful mess

There is a fourth, and it explains why small talk stalls where it does.

Anna Bruk, Sabine Scholl and Herbert Bless published work in the Journal of Personality and Social Psychology in 2018 on what they called the beautiful mess effect. When people imagined showing vulnerability — admitting a mistake, confessing an uncertainty, asking for help — they rated it as weak and messy. When they watched someone else do the same thing, they rated it as courageous and appealing. The self-view and the other-view diverged sharply.

This is why so many exchanges get stuck at the level of the weather. Not because nobody wants to go further, but because each person believes that going further is a risk they would be taking alone.

Kardas, Kumar and Epley extended this in the same journal in 2022, showing that people expect deeper conversations to be more awkward than they turn out to be, and consistently underestimate how much the other person will care about what they say. The reliable finding across this literature is that the conversation you are avoiding is usually better than the one you settle for.

What small talk is actually doing

The standard complaint is that small talk is meaningless. I want to argue that the complaint has the function wrong, in the way that calling a handshake meaningless has the function wrong.

Three things are happening under the surface of a dull exchange about traffic.

It establishes safety before content. Nobody opens with the thing that matters. The weather is a low-stakes signal test: I am willing to spend attention on you, are you willing to spend attention on me. If that exchange goes well, a door opens. If it goes badly, nothing was lost. Every deeper conversation you have ever had was preceded by one of these, whether or not you noticed.

It maintains the weak ties. Mark Granovetter's 1973 work on the strength of weak ties made the case that acquaintances — not close friends — are the ones who carry novel information into your life, because your close friends already know what you know. The barista, the neighbour, the person two desks over. These relationships are maintained almost entirely through small talk, and they degrade to nothing without it.

It is a low-dose antidote to a specific loneliness. Not the ache of having no one close. The other kind: moving through a day surrounded by people and being addressed by none of them. That one responds surprisingly well to being recognised for eleven seconds by someone whose name you may not know.

I notice this most at the ordinary edges of a day — the school pickup, the queue, the neighbour with the recycling bin. There is nothing in any of it. And a week with a dozen of those small exchanges feels different from a week with none, in a way I could not have predicted and cannot really argue with.

Making it less draining without faking enthusiasm

Everything above is a case that small talk is worth doing. None of it is a case that it is free. For a lot of people, and I would include myself, it genuinely costs energy. So the question is not how to love it. It is how to lower the price.

Stop performing and start being curious. The exhausting version of small talk is the one where you are monitoring your own output — being interesting, being funny, not going quiet. Curiosity moves the attention off you, which is where the drain was. It is also far more effective. Nobody has ever walked away from a conversation thinking the other person listened too carefully.

Ask the second question. The whole difference between hollow and warm is one follow-up. "Busy week?" — "Yeah, we're moving house." — and then either "Oh, nice" and it dies, or "Where to?" and you are suddenly in an actual conversation. The second question costs nothing and signals that the first one was not a formality.

Notice something instead of asking something. Questions can feel like an interview. Observations invite without demanding: "That queue is not moving at all." Anyone can pick it up or let it go, and neither is a rejection.

Give it a shape. A conversation with no visible end is what most people are actually dreading. "I've got two minutes before my call" is not rude, it is a container, and it makes both people relax.

Skip the enthusiasm. You do not need to be bright. Bright is a performance and it is expensive. Attentive is enough, and attentive is what people remember.

Let it be short. Ninety seconds is a complete interaction. The instinct to keep going until it reaches some threshold of substance is what makes it tiring, and it is not required.

Where the line is

I want to be careful here, because "talk to more people" is advice that lands badly on people who have good reasons not to.

Some avoidance is just accurate self-knowledge. If you have four hours of social contact behind you and your capacity is spent, declining the fifth is management, not a failure. Introversion is a real difference in where recovery comes from, not a defect to be corrected. And there are people and settings where the exchange genuinely is unpleasant — that judgment is usually right, and you should keep trusting it.

Two markers distinguish healthy avoidance from the habitual kind.

The first is whether you are choosing or defaulting. Healthy avoidance is a decision you can describe: I am tired, I have nothing left today. Habitual avoidance has no reasoning in it. The phone is out before the lift doors close, every time, regardless of how you feel that morning.

The second is what happens afterwards. Protecting your energy leaves you feeling protected. Habitual avoidance tends to leave a small residue — a flicker of relief, then something flatter underneath. If you consistently feel slightly worse after skipping the interaction than you predicted you would feel after having it, the prediction is the thing that is broken, not you.

Habitual isolation also has a mechanism that keeps it running. Skills you do not use get rusty, rusty skills make the next attempt feel harder, and the harder it feels, the more sensible avoidance seems. It tightens quietly. That is worth knowing not as a reason for alarm, but because the loop runs equally well in reverse and does not require much to turn it.

Three low-effort prompts to try this week

Not a challenge. Three, spread across a week, is enough to get a reading.

One. The specific compliment to a stranger. Not "nice jacket" as a reflex — something you actually noticed. It takes four seconds, it is nearly impossible to receive badly, and it requires nothing further from either of you. Good starting point if the whole idea makes you tense.

Two. The name and the follow-up. With someone you see regularly but have never gone past a nod with, ask their name, and next time use it and ask one thing. This is the single highest-return move on the list. A person who uses your name has moved from stranger to acquaintance in one step, and the entire weak-tie layer of your life is built out of exactly this.

Three. Ninety seconds before the business. On a call or a meeting, before the agenda, ask one real question and listen to the answer. Not "how are you" answered with "good". Something with an actual object in it: how was the trip, how did the thing on Friday go. Most professional relationships never get this, and they are noticeably worse for it.

Then notice one thing afterwards: how it went compared with how you expected it to go. That comparison is the experiment. If the research holds, you will find the gap sitting exactly where it has been sitting for everyone else.

Questions people actually ask

I am an introvert. Does this apply to me? The biases do — they show up across personality types. What changes is the budget. Introversion means social contact draws down a reserve that takes quiet to refill, so the useful version of this is not "more talking" but "fewer avoided moments that you would actually have enjoyed". Shorter and more frequent tends to work better than long and rare.

What about social anxiety? The liking gap is larger in people who are more socially anxious, which is to say the misprediction is bigger, not that the anxiety is imaginary. Behavioural approaches to social anxiety often work precisely by testing these predictions in small, survivable doses. If avoidance is meaningfully shaping your life, that is worth working on with a professional rather than a list of prompts.

Isn't small talk dishonest? A handshake is not a lie about how much you like someone. Small talk is a protocol — it carries almost no content and it is not supposed to. The dishonesty people object to is usually the enthusiasm they think is required, and that part is genuinely optional.

What if the other person clearly does not want to talk? Then stop, easily and without reading anything into it. Most people are receptive far more often than we expect, but some are tired, late, or having a bad day. A brief exchange that does not land costs you approximately nothing, which is the underrated part of the whole enterprise.

Does any of this help with real loneliness? Partly, and it is worth being honest about the limits. Weak ties address one kind of loneliness — the unrecognised, unaddressed feeling of moving through a populated day alone. They do not substitute for close relationships. But close relationships almost always begin as weak ties, so the layer is not separate from the thing you want. It is where the thing you want comes from.