Why does cutting one wire sometimes take down the whole internet, and sometimes nothing at all?
After you watchWhy does cutting one wire sometimes take down the whole internet, and sometimes nothing at all?
The short answer
Cutting one wire takes down a network only when that wire is the single path holding a chunk of it together. If a network has more than one path between its points (a mesh or web), one cut just reroutes and nobody notices; if it has a single chain or one central hub, the right cut splits it and part goes dark.
Try this next
- What if you cut two wires in the web instead of one? Predict whether two snips can finally split the web, then make two cuts and watch — does it depend on which two you pick?
- What if you snip right at the hub of the single-road network? Guess how much goes dark from one cut at the center versus one cut out on a spoke, then try both and compare.
Now you — bend it
- What if What if you add just one extra wire to the single-road network — can you make it survive a cut?A second path between two points means a cut on one still leaves a detour. Where would one wire help most?
- What if What if you remove wires from the web until one snip finally splits it?You're hunting for the moment a pair of points drops to just one path — that's when a bridge appears.
Can you prove it?A web survives one cut because every two points have at least two separate paths. — Pick any two dots in the web, trace one path between them, then trace a second path that shares no wires with the first. Cut a wire on one path and check the message still gets through the other.
Design your own test:Before you cut, predict exactly which dots will go dark — then snip and see if you guessed the isolated chunk right.
Explain it to a 6-year-old: One road into town: block it and you're stuck. Lots of streets: block one and you just go around.
The whole story
How it works
A network is dots (computers) joined by lines (wires), and a message travels along a path of wires from one dot to another. In a single-road network — a line, or everything hanging off one hub — there is exactly one path between two points, so snipping any wire on that path leaves the far side with no way through. In a web-of-streets network, every dot connects to several others, so between any two points there are several different paths; cut one wire and the message simply takes another route. You'd have to cut many wires before anyone is truly isolated.
What people get wrong
People think a network is reliable just because it has lots of wires in total. What actually matters is having more than ONE path between every two points. A hub-and-spoke network can have tons of wires yet still die from one snip at the hub, because every path runs through that single weak link.
The catch
The web of streets survives cuts because of its spare paths, but those extra wires cost real money to build and maintain — you're paying for backups you hope you never need. The single road is the cheapest way to connect everyone, but it has a weak link: one bad cut (or one failed hub) and a whole chunk goes offline with no detour.
Questions kids ask
So is the internet a chain or a web?
Mostly a web. The internet was designed with many redundant connections so that if one cable or router fails, traffic automatically reroutes around it. That redundancy is the whole reason a single cut usually goes unnoticed.
If a web is safer, why do some places still get knocked offline by one cut?
Because not every part of a network is a web. The edges — a single town, an island, or a building on one cable — often hang off a single path or one hub. That spot is a weak link, so one cut there takes everything past it offline.
Does adding more wires always make a network safer?
Not by itself. Piling wires onto one central hub still leaves every path running through that hub — cut it and everything fails. Safety comes from having more than one separate path between points, not just from a bigger pile of wires.
How many wires would you have to cut to break a web network?
More than one — that's the point. In a web, every two points have at least two separate paths, so you have to cut every one of those paths before a part is isolated. The more separate paths there are, the more cuts it takes.
Talk about it
- Guess: how many roads into our town are there, and what happens if one is closed?
- If you were drawing a network so it never fully breaks, where would you add a spare path first?
- Why might someone build the cheaper single-road network even though they know it can fail?
For grown-ups
This is graph theory. Reliability comes not from total edge count but from connectivity — how many edges or nodes you must remove to disconnect the graph. An edge whose removal splits the graph is a bridge; a node that does so is an articulation point (a hub center is one). A mesh has no bridges: every pair of nodes is joined by at least two edge-disjoint paths (related to Menger's theorem), so any single cut leaves a detour. The early internet, ARPANET, was deliberately built with redundant routing for exactly this fault tolerance; the cost of that redundancy is paying for spare links that sit idle until something fails.
Keep going
What else makes you wonder?
- If one hub is a weak link, how does anyone decide where to put it?
- Your body has nerves and blood vessels — are those built like a chain or a web?
- Could you make a network too tangled, with so many spare paths it's wasteful?