spread & inequality · metaphor 49 of 100
Your close friends already know what you know. The acquaintance you almost didn't message — the college roommate's cousin, the stranger from the conference — is a bridge to a world your whole circle can't see. Weak ties carry the news, and a handful of them make the planet six steps wide.
When Mark Granovetter asked professionals how they had found their jobs, the answer refused to flatter friendship: of those who heard through a personal contact, most heard it from someone they saw occasionally or rarely — the former colleague, the friend of a friend — not from anyone they would have called close. Apartments travel the same way. So do ideas worth stealing, and a startling number of spouses. The people who love you most are, reliably, the last to know something you don't.
The reason is structural. Friendship clusters: your friends are friends with each other, which is exactly what makes them warm and exactly what makes them redundant. What circulates inside the circle is your own information, coming back around. The rare long link that leaves the circle is worth more than another strong tie inside it — and Duncan Watts and Steven Strogatz found the deeper shock: it takes only a few such links to collapse a big, cliquish world into a small one.
The rewiring dial · Watts–Strogatz, live
Sixty people stand in a ring, each tied to their four nearest neighbors — the village. Drag p to rewire a fraction of those ties to somewhere random. Two numbers are recomputed for the fresh graph at every notch: L, the average shortest path between everyone (how far you are from arbitrary information), and C, the clustering coefficient (how often your friends are friends with each other).
honest computation — every notch generates a real graph; L is exact breadth-first search over all pairs, C the exact clustering coefficient. curves = mean of 10 fresh graphs per notch; dots = this graph, right now. window shaded where half of L's collapse is already done while ≥70% of C survives.
the strength of weak ties
Look at the village end of the dial. C = 0.5: half of all possible friendships among your friends actually exist. That's the comfort of a strong-tie world — and its trap. Anything your best friend learns, you were going to hear from two other people anyway. A strong tie added inside the cluster buys trust but almost no new reach: its information is already yours, recirculated.
A rewired tie is different in kind, not degree. It is the only path between your world and another; nothing else you have duplicates it. The bridge's value is its reach, not its warmth. Below, each node's information is painted by where it lives on the ring. Watch how little your neighbors' colors differ from yours — and how the one contact across a long tie glows.
The job lead · novel information per contact
Pick who you are. Each contact's novelty is the color-distance between their information and yours, computed on the live graph — 0% means they know what you know.
this graph: the 60-node village with 3 ties rewired to somewhere at least a third of the ring away (kept long for legibility; readouts are computed, not set). click any node, or slide.
Every node here has 3 to 5 ties — nobody is popular. A small world is what a handful of long ties does to a flat, egalitarian network. Networks where a few nodes hoard thousands of ties are a different machine with different consequences — that story is preferential attachment. Shortcuts shrink distance; hubs concentrate it.
what to try
Drag the dial slowly up from p = 0. The blue curve (distance) falls off a cliff while the gold curve (warmth) barely moves. In the shaded window the graph still feels like a village to everyone in it — and is already, measurably, a city.
Below: the same secret, seeded at the same person, in the village and in the village plus two rewired ties. Then tick "strong ties only" and watch the small world go back to being a large one.
Slide through the job-lead panel. Most nodes have no bridge — every contact under 10% novel. A few have exactly one. Notice which kind of node you'd rather be when the graph carries a job opening.
Press Show the longest tie and read what cutting it would cost the whole graph. Then the real exercise: who is your longest tie in life — the person reaching a world none of your other ties reach — and when did you last message them?
six degrees, few shortcuts
The strangeness in the chart above is how little it takes. Two rewired ties out of a hundred and twenty already pull everyone in this graph about a quarter closer to everyone else — and the village doesn't feel it: clustering barely moves. Rewire one tie in ten and the average distance falls by nearly half while three quarters of the warmth still stands. Warmth degrades only in proportion to what you rewire; reach improves out of all proportion to it. That asymmetry is the theorem, and it is why Stanley Milgram's letters, forwarded acquaintance to acquaintance from Nebraska, could land on a Boston stockbroker's desk in about six hops: the planet doesn't need many bridges to be small, and it has more than a few.
Read it at the scale of one life: you do not need to know many kinds of people. One cross-country friendship, one colleague in a different industry, one mentor from another generation — each is a rewired edge, and each one collapses your effective distance to everything on the far side of it. The village never has to stop being a village. It just needs a door.
The rumor race · one hop per tick
Same seed, same rule — each tick, everyone who knows tells everyone they're tied to. Click any node to move the seed.
The villagep = 0
—
+ two shortcuts2 of 120 ties rewired
—
honest computation — the wavefront is real breadth-first search on both graphs; tick counts are measured, not scripted. shortcuts are rewired to somewhere at least a third of the ring away.
tending your bridges
Strong ties are self-repairing: miss a friend's call and the cluster routes around it — you'll hear their news anyway, and see them at the same table next month. A bridge has no cluster behind it. Nothing reminds you of that person; no third friend mentions them; if the tie lapses, no structure regrows it. The most valuable edges in the graph are precisely the ones that decay silently, and they do not form by accident either. They form at rewiring events — the move, the conference, the odd hobby, the class taken for no good reason — moments when you're briefly tied to a ring that isn't yours.
So run the audit the instrument suggests: of everyone you talk to, how many reach a cluster none of your other ties reach? For most people the honest count is one or two — and those one or two get less maintenance than the fifth-closest friend, because nothing in the week naturally routes through them. Structurally, that is investing everything in redundancy and starving the only non-redundant edges you own. A message costing five minutes keeps a world attached to yours.
the mapping
| Mathematics | Life |
|---|---|
| the ring lattice | The village: warm, tightly clustered, and far from everything it can't see. |
| clustering C | Your friends knowing each other — comfort, trust, and redundancy in one number. |
| path length L | How far you are from arbitrary information: the job not posted, the idea not published yet. |
| a rewired edge | The odd acquaintance across worlds — the weak tie whose value is reach, not warmth. |
| the gold window | Village warmth with city reach: almost all strong ties, plus a few deliberate strange ones. |
| the bridge | The tie whose loss doesn't cost you a friend — it disconnects you from a world. |
where the metaphor tears
Bridges move information: the lead, the idea, the introduction. Moving anything heavy — money, childcare, a reputation staked on your behalf, a 3 a.m. phone call — still requires strong ties, because those transfers run on trust that only clustering builds. The two currencies are different, and a life optimized for reach alone is as poor as one optimized for warmth alone. You need the village and the door.
Every node here has about four ties; reach is cheap and equally purchasable. Real social worlds also have hubs, gatekeepers, and homophily — bridges form more easily for people who already resemble the far cluster, and some rings are walled. The Watts–Strogatz model explains why a few links suffice; it says nothing about who gets to have them. For the inequality half of the story, see preferential attachment.
The audit is for noticing structure, not for networking sociopathy. People detect being treated as an edge in someone's graph, and the tie dies of it — the bridge's bandwidth was the mutual, unforced interest. In practice the durable long ties form as byproducts of genuinely odd enthusiasms: you cannot farm serendipity, but you can keep living in places where it grows.