Take a rubber band. Stretch it, twist it, squish it into any shape you like, as long as you don't cut it or glue two ends together. Is it still "the same" rubber band?
Every distance inside it just changed. The part that used to be an inch from your thumb might now be three inches away. So in what sense is anything the same?
In the sense that matters to a topologist: nothing tore, and nothing got glued together that wasn't already touching. That's the entire subject. Topology studies the properties of a space that survive stretching, bending, and squishing, and throws away the ones that don't, like distance, angle, and area.
If stretching destroys every distance, what's left to measure?
Picture five beads threaded on a rubber band: A, B, C, D, E, in that order. Stretch the band unevenly, more on one end than the other, and every gap between beads changes by a different amount.
Stretch the band unevenly. The beads never swap places, even though the gaps between them stop matching.
Drag the slider and watch the gaps warp. The distance from A to B and the distance from D to E start out equal and pull apart fast. But notice what never happens, no matter how hard you stretch: the beads never pass through each other. A is always left of B, which is always left of C. Order survives. Distance doesn't.
That's the whole shift in perspective topology asks you to make. Forget "how far apart," and ask "what's next to what." A rubber band deformation can scramble every number you'd get from a ruler, but it can't scramble that kind of neighboring relationship, unless it actually tears the material.
A coffee mug and a donut: is that a joke or real math?
You've probably heard the joke: a topologist can't tell their coffee cup from their donut. It's not really a joke. It's the single clearest example in the whole field.
A coffee mug is a solid blob of material with one hole through it, the handle. A donut is a solid ring with one hole through the middle. Different shape, same hole count. And a hole count is exactly the kind of thing that survives stretching.
Press play. The handle slides inward and swells into a donut hole. Nothing is ever cut or glued.
Press play. Watch the handle's hole slide toward the center of the blob and swell in size until it becomes the donut's hole. At every single frame of that animation, the shape has exactly one hole. Nothing gets cut to make room for the hole to move, and nothing gets sealed shut. The mug and the donut are, topologically, one object shown at two different moments of the same stretch.
Compare that to the plain disk drawn beside it. A disk has zero holes. No amount of squishing turns a zero-hole shape into a one-hole shape, short of tearing it open, and tearing isn't allowed. That's why a topologist really can tell a donut from a plate, even while insisting a donut and a mug are the same thing. The count of holes is the invariant. The exact shape around them is just decoration.
We have a name for two shapes related by this kind of stretch-no-tear deformation: they're homeomorphic. We write to say and are homeomorphic, and it means there's a way to continuously deform one into the other, with a continuous way back, matching every point to exactly one point on the other side.
How do you actually count a hole, without just eyeballing it?
"One hole" is easy to see in a picture. It's a lot harder to check by formula, especially for a shape you can't draw, like a space with a hundred dimensions. Topologists needed a test that doesn't rely on looking.
Here's the trick: draw a loop on the surface, and try to shrink it to a single point without ever leaving the surface.
Drag the slider to shrink the loop toward a point. Switch surfaces to see when it gets stuck.
On a solid disk, drag the slider and the loop shrinks all the way down, no matter where you started it. There's nothing in the way. But switch to the disk with a hole punched out, and try again. The loop shrinks fine at first, then jams right at the edge of the hole. It physically cannot pass through space that isn't there anymore. The loop gets stuck exactly because a hole exists, and that stuck-ness is a fact you can check without ever measuring a distance.
This loop-shrinking test is the seed of an entire tool called the fundamental group, which tracks every fundamentally different way a loop can sit on a surface. A disk's fundamental group is trivial, every loop shrinks. A donut's is not, loops that go around the hole or through the middle can never shrink away. Counting holes by eye is a shortcut. Testing whether loops shrink is the actual rigorous version of that shortcut.
If distance doesn't matter, what does topology actually track?
Strip away distance, angle, and shape entirely, and you're left with one thing: which points are connected to which. That's not an approximation of topology, that's close to the whole definition.
Drag any node anywhere. The connections don't care where the points sit, only cutting one does.
Drag any node anywhere you want. The distance between A and E on your screen will swing wildly, but the number of connected pieces never budges, it stays at one, because you haven't touched a single connection. Now click "cut edge C–D." You haven't moved a single node. But the graph instantly splits into two pieces, because that's the one thing that actually matters here. Repositioning is invisible to topology. Cutting or gluing a connection is the only thing it can see.
This is why topologists talk about "open sets" instead of distances. We write a topology on a set as a collection of subsets of , called the open sets, satisfying three rules:
Notice what's missing from that list: numbers. No formula for distance appears anywhere. All a topology gives you is a rule for which clusters of points count as "close enough to surround a point," and that's already enough to talk rigorously about stretching, connectedness, and holes, with no ruler required.
Why does any of this actually matter outside a math class?
Subway maps are the everyday version of this idea already. The London Underground map is wildly wrong about actual distance and direction between stations, and nobody cares, because all a rider needs is which line connects to which. A subway map is a topology, drawn on purpose.
The same shift shows up everywhere once you know to look for it. A computer network doesn't care about the physical distance between two servers, only whether a path of connections exists between them. Robotics leans on it too: the space of every angle a robot's joints can bend into isn't flat, it wraps around, and knowing its topology tells engineers which configurations are reachable from which without ever computing a single distance. Data scientists use a whole field called topological data analysis to find holes and clusters in messy, high-dimensional data where "distance" barely means anything sensible to begin with.
A subway rider ignoring geography, a robot arm ignoring exact angles, and a coffee mug pretending to be a donut are all leaning on the same idea.
The short version
Topology studies whatever survives continuous stretching and bending, and throws out everything that a rip or a fresh glue-job would be needed to fix. Distance and angle don't survive, so topology ignores them entirely. Connectivity does survive, so does the count of holes, and those become the real objects of study. A coffee mug and a donut count as the same shape because one continuously deforms into the other. A loop that can't shrink to a point is proof a hole exists, checkable without a ruler. And a topology, formally, is just a rule for which sets of points count as "open," with no notion of distance built in at all.
Next time someone hands you a subway map, remember: you're already reading a topology, not a geography.
All visualizations are interactive React components running entirely in your browser, using SVG. No libraries beyond React.