There is a specific kind of evening that has become unremarkable. Six people in a room who genuinely like each other, all present, all technically together, and all looking down. Nobody is being rude. The conversation ran out and the phone is the path of least resistance.
A game fixes this faster than anything else available, and it is worth being precise about why — because the reason is not nostalgia, and it is not that games are inherently fun. It is that a game is a structure, and structure is the thing that has gone missing.
The number underneath all of this
An analysis of two decades of American Time Use Survey data, published by Viji Kannan and Peter Veazie in SSM - Population Health in 2022, tracked how Americans actually spend their social hours. Two of its figures are worth sitting with.
Average time spent with friends fell from about 60 minutes a day in 2003 to about 20 minutes a day in 2020. Over the same period, average time spent alone rose from 285 minutes to 333.
The obvious explanation is the wrong one. 2020 was a pandemic year and it is tempting to read the whole thing as lockdown, but the decline was already well underway — the paper finds the fall in time with friends running through the 2014–2019 stretch, before anything closed. Whatever is happening here started long before it had an excuse.
The public-health framing arrived in 2023, when the US Surgeon General issued an advisory on loneliness and isolation treating social connection as a health issue rather than a lifestyle one. That is the context in which "we should play a game" stops sounding trivial.
What actually went missing
It is tempting to read a two-thirds fall in time with friends as a story about phones, and phones are certainly in it. But the shape of the decline suggests something broader, because it runs steadily through a period in which the specific technologies changed several times over.
The more mundane account is that the occasions went first. Time with friends is mostly unplanned time — the hour after something else finishes, the evening nobody organised. That kind of time depends on a set of conditions that have quietly eroded: living near the people you like, having a regular place you both end up in, and having an evening whose end is not already spoken for.
When unplanned time disappears, what remains has to be arranged, and arranged time behaves differently. It gets scheduled weeks out, it is easier to cancel, and — the part that matters here — it arrives with an expectation that it will be worth the effort of having organised it. An evening that took three weeks to coordinate carries a quiet pressure to be good.
That pressure is precisely what makes an unstructured group evening prone to stalling. Six people who see each other constantly can sit in comfortable silence. Six people who cleared a Saturday for this cannot, and everyone can feel it.
Attention needs a reason
In an unstructured group conversation, paying attention is a choice each person makes continuously. It costs something, the cost is invisible, and the moment it exceeds the perceived value someone reaches for their pocket. Nobody decides to disengage. It just becomes marginally easier than not disengaging.
A game inverts that default. Turn order gives your attention a deadline. A shared prompt means everyone is looking at the same object. A score means what just happened has consequences, however trivial. None of that is about the content of the game — a group can play something fairly stupid and still get the effect, because the effect is coming from the structure rather than from the material.
The game is rarely the point. The reason to look up is the point.
What the phone research does and does not show
Here is where most articles on this subject overclaim, so it is worth being careful.
The study everyone cites is Przybylski and Weinstein (2013), which found that the mere presence of a phone on the table — not being used, just present — lowered how close and trusting two strangers felt after a conversation. It is a compelling result and it has been repeated in a thousand articles as established fact.
It has also repeatedly failed to replicate. A PLOS One replication published in 2021 did not reproduce the effect, and notes it is not alone in that. The honest summary is that mere presence is a much weaker and less reliable finding than its popularity implies.
What holds up better is the research on phubbing — actually using a phone while someone is talking to you. That has consistent support: people who are phubbed report lower conversation quality and a reduced sense of belonging, which is roughly what anyone would predict.
Why include this at all? Because the weaker version is still enough to make the point, and because a page that cites the convenient study while ignoring the failed replications is not worth trusting on anything else it says.
So: the case for group games does not rest on phones being uniquely corrosive. It rests on something duller and better established — that in the absence of any structure, groups drift, and that the drift has an easy destination.
The constraint is the active ingredient
Ask a group "so what has everyone been up to" and you get four sentences and a lull. Ask the same group something specific and slightly unreasonable — what is the pettiest reason you have ever stopped talking to someone — and you get twenty minutes.
The second question is not more interesting than the first. It is narrower. An open question makes each person responsible for choosing what to say, which is work, and under mild social pressure people default to the safest available answer. A narrow question removes the choice, and the safe option disappears with it.
That is the mechanism behind nearly every social game that works. Truth or Dare is a constraint generator. Would You Rather is a constraint generator. Never Have I Ever is a constraint generator with scoring bolted on. The theme is packaging; the constraint is the product.
It also explains why games work better than the sincere alternative. "Let us all put our phones away and actually talk" asks people to generate the content and the willingness at the same time. A prompt supplies the content, so all that is left to supply is the willingness.
Not every constraint works, though, and the ones that fail do so in recognisable ways. A useful constraint has three properties, and dropping any of them produces a specific kind of dead round.
- It has to be answerable immediately. If a question requires thinking, the room waits, and waiting is where a game loses its momentum. The best prompts have an answer that arrives before the person has decided whether to give it.
- It has to have more than one defensible answer. A question with an obvious correct response is a quiz, not a conversation. The interesting prompts are the ones where two reasonable people go in different directions and then have to explain themselves.
- It has to be equally answerable by everyone present. A prompt that assumes a shared reference, a relationship or a particular decade excludes whoever lacks it, and being excluded once is usually enough to stop somebody volunteering for the rest of the evening.
The third is the one that gets violated most in mixed groups, and it is almost always accidental. A question about university is not intended to exclude the two people who did not go — but it does, and they will notice a good deal faster than whoever asked it.
Why "let us all put our phones away" does not work
The sincere alternative to a game is to name the problem out loud. Somebody says it, usually kindly, and it produces about four minutes of dutiful eye contact before the room returns to exactly where it was.
It fails for a reason worth understanding, because the reason is also the argument for games. Asking a group to put their phones away requires them to generate the willingness and the content at the same moment. You have removed the easy option without supplying anything to replace it, so the group is now sitting in the same conversational vacuum with fewer exits.
A prompt supplies the content. That is the entire difference. Whatever is left for the group to supply is only the willingness, which is a much smaller ask and one that people are generally happy to meet.
There is a second thing going on. Naming the problem makes it somebody's fault — usually whoever was on their phone when the observation was made. Suggesting a game makes it nobody's fault, because the game is a change of activity rather than a correction of behaviour. The social cost of proposing it is close to zero, which is why it is the intervention that actually gets used.
Almost nobody resents being handed a question. A surprising number of people resent being told to pay attention.
Why some games survive and others evaporate
Some games end and vanish. Others leave something behind — a score, a result screen, a ranking, a story that gets retold. The second kind gets requested again, and that is more predictive than how enjoyable the game was while it was happening.
The artefact does two jobs. It gives the evening a shape, so it can be referred to afterwards as an event rather than as a blur. And it gives somebody a reason to send it to a person who was not there, which is the only way a game escapes the room it was played in.
Four properties show up in the games that get asked for a second time:
- They start in under a minute, explanation included.
- Every player has something to do on every round, not only on their turn.
- They have a natural stopping point, so the group ends on a high instead of watching it die.
- They produce something shareable, even if it is only a number.
Missing one of the four is survivable. Missing the first is not — a game that takes five minutes to explain does not get a second outing regardless of how good it is once running.
It is a useful filter to apply to a shelf of board games, incidentally. The ones gathering dust are almost never the bad games; they are the ones with a fifteen-minute setup, which means they need an evening planned around them rather than an evening they can be dropped into.
What this looks like on an actual evening
All of the above is abstract, so here is the concrete version. Nothing in it takes more than a minute to start.
The lowest-effort intervention is a question with a forced choice — Would You Rather being the canonical form. It requires no explanation, no turn order and no disclosure, which makes it the safest thing to reach for when you are not sure whether the room wants a game at all. If it does not land, nothing has been lost, because nobody had to commit to anything.
One step up, where the group knows each other, are the formats built around prediction: a quiz about one person that everyone else answers, or the two-person version where a pair guess each other separately and the answers get compared. These produce the artefact discussed above almost automatically, because a score is generated whether you wanted one or not.
For an evening rather than twenty minutes, the games with a genuine escalation curve are the ones that hold — Truth or Dare being the obvious one, though it needs more preparation than its reputation suggests. Choosing prompts that suit the specific group matters more than the game choice itself, and the hosting run sheet covers the ordering across a whole night.
If you want the full list with player counts and running times, twenty games that need nothing but a phone is sorted by how fast each one starts, which is the variable that turns out to matter most.
The honest limitation
None of this makes a group closer by itself. A game is a structure for attention, and attention is necessary but nowhere near sufficient. What actually moves a friendship is what people choose to say once the structure has given them the opening, and no prompt bank controls that.
There is also a real ceiling on what twenty minutes of Would You Rather can do about a two-decade decline in time spent with friends. Anyone claiming otherwise is selling something.
What a game can do is make the opening reliable. On an evening that would otherwise have been six people and six screens, that is not nothing — and given how much less of that time there now is, making the remaining hour count is a more reasonable ambition than it sounds.
