Why does one person staying home from a party tip it from packed to empty?
After you watchWhy does one person staying home from a party tip it from packed to empty?
The short answer
Because people decide by watching each other. When everyone only comes if enough others are coming, a packed party can be balanced right at a tipping point. Take one person away and the headcount drops below the shyest guest's number, so they leave too, which drops it below the next guest's number, and so on — a chain reaction empties the whole room. One person can flip the whole crowd.
Try this next
- What if you fill the room with only easygoing, low-number people? Set everyone's number low, predict whether one person leaving still empties the room, then run it and watch how far the dominoes fall.
- What if you add one person instead of taking one away? Park the crowd at its edge, predict the new headcount, then drop in one easygoing guest and watch an empty room snap toward packed.
- What if the same crowd ignores everyone (every number zero)? Set all the numbers to zero, predict what removing one guest does, then run it and see if the cliff is gone.
Now you — bend it
- What if Give everyone in the room a high number so they're all shy. Predict the cascade size, then take one guest away.When the shyest people are stacked near the edge, one leaver knocks the count below the next, then the next — the dominoes can run all the way down.
- What if Mix a few easygoing low-number people in with the shy ones, then remove one guest.The easygoing ones stay even when the count dips, so they can stop the chain partway and the room only half-empties instead of clearing.
- What if Run it in reverse: start near empty and drop in one easygoing person.If their arrival pushes the count past the next person's number, that person comes too, and a quiet room can snap toward packed.
Can you prove it?Removing one person can empty the whole party, not just leave one fewer. — Park the copycat crowd right at its edge and write down the headcount. Take one person away and watch — the count drops far past one. Then set every number to zero and take one away again: this time you really do get just one fewer. The cliff only shows up when people copy each other.
Design your own test:Before you press play, predict: with the numbers you picked, will taking one guest away clear the room, half-empty it, or barely change it?
Explain it to a 6-year-old: Everybody copies their friends, so when one kid goes home, the next copies them, then the next — and soon the whole party leaves together.
The whole story
How it works
Give each person a number: the smallest crowd they'll join. Some are easygoing (a low number), some are shy (a high number), and everyone keeps checking the headcount. The party settles where the number of people willing to come at that headcount matches the headcount itself. Right at the edge, that balance is fragile: removing one guest pushes the count just below the shyest person's number, they leave, which pushes it below the next person's, who leaves — the dominoes fall all the way down. A crowd where nobody watches anyone (everyone's number is zero) has no edge at all: take one away and you simply have one fewer.
What people get wrong
People assume a small change to the input gives a small change to the result — that taking one guest away should leave one fewer guest, smoothly. But when choices feed back on choices, outcomes don't drift, they tip. Near a tipping point one person changes everyone, so the result jumps from packed to empty all at once instead of dropping by one.
The catch
A crowd that copies each other can catch fire: a tiny spark snowballs into a packed room incredibly fast, which is how a craze takes off overnight — but it's fragile and hard to predict, since the same crowd can crash to empty just as fast off one person. A crowd that ignores everyone is rock steady and predictable, so you always know roughly how many will show — but it can never snowball or catch fire; it just plods along, one in, one out.
Questions kids ask
How can one person empty a whole party?
Because everyone is watching everyone. If the party is balanced right at the edge, taking one person away drops the count below the shyest guest's number, so they leave. That drops it below the next person's number, who leaves too, and the dominoes fall all the way down until the room is empty. The one person didn't leave alone — they started a chain reaction.
Wouldn't taking one person away just leave one fewer guest?
Only if nobody copies anybody. If everyone comes 'no matter what', then yes — one fewer. But if everyone only comes when enough others are there, the crowd has a cliff near the edge, and one person leaving can tip everyone out instead of just one.
What other things tip like this?
Standing ovations (one person stands, then everyone does), fashions and crazes catching on overnight, and bank runs, where a few people pulling their money out scares the next people into pulling theirs. They all work because each person reacts to what the others are doing, so the crowd snaps all-or-nothing instead of changing a little at a time.
Does the same idea work in reverse, from empty to packed?
Yes. Near the edge, adding one easygoing person can push the count past the next person's number, who then comes, which pushes it past the next, and a quiet room snaps to packed. Tipping points work both ways: a tiny nudge can flip the whole crowd up or down.
Talk about it
- Guess first: if 30 people are at a party balanced on the edge, how many leave when one goes home?
- Where have we seen a whole room copy each other all at once — dancing, standing up, leaving?
- Which one person at our table do you think everyone else is secretly watching to decide what to do?
For grown-ups
This is Mark Granovetter's threshold model of collective behavior (1978). Each person joins once the number already participating crosses their personal threshold, so choices feed back on choices. The settled crowd sits where the cumulative threshold curve meets the 45° line; near a tangency that equilibrium is unstable, so an arbitrarily small perturbation triggers a cascade to a far-away equilibrium — a tipping point, or phase transition. The same threshold-and-feedback math underlies bank runs, riots, standing ovations, fads, and how behaviors spread through social networks. With no feedback (every threshold zero) there is no cliff: attendance just tracks the headcount smoothly.
Keep going
What else makes you wonder?
- How big does a crowd have to be before one person can't tip it anymore?
- If the room can be packed or empty from the same start, what decides which way it falls?
- Could you guess in advance which one person is the one holding the whole party together?