A one-ended hyperbolic group --- one whose large-scale geometry resembles negatively curved space and that cannot be disconnected at infinity --- can sometimes be split in two by removing a surprisingly thin subset. The question is how thin. The answer, established through quantitative coarse geometry, is precise: a one-ended hyperbolic group that is not a surface group admits a cut of subexponential growth if and only if it splits over a virtually cyclic subgroup. The group's algebraic decomposition is exactly characterized by the volume growth rate of its thinnest separating sets.
The “only if” direction carries the sharper surprise. If the group does not split over a virtually cyclic subgroup, then every separating subset must grow exponentially --- there is no thin way to divide it. Sufficiently large thickened spheres in such groups are hard to cut in the sense that any cut-set through them must be exponentially large. The group's resistance to separation is not just a topological fact but a quantitative one, measurable in growth rates.
This connects algebraic structure to geometric bulk in an unusually direct way. Group splittings are algebraic objects --- they describe how a group decomposes as an amalgamated product or HNN extension. Separation profiles are geometric objects --- they measure how much material must be removed to disconnect a space. The theorem says these are the same information, encoded in different languages.
The broader lesson is that in any structured system, the cost of division reveals the nature of the structure. When a system can be cheaply separated, it was never truly unified --- it was two things joined by a thin bridge. When every cut is expensive, the integration runs deep. The price of partition is a diagnostic for the depth of connection.