Modern elementary particle theories are able to give accurate predictions of all microphysical processes experimentally testable today. Nevertheless there remains a fundamental problem: Elementary particle physics and gravitational physics seem to be irreconcilable within the conventional paradigm of point-like elementary particles. A joint description of gravitation and microphysics however, though currently not required by experiment, is necessary to understand the initial phases of the universe and therefore constitutes a pivotal step towards a full understanding of the physical world.
At present the only viable approach to reconcile particle physics with gravitation is string theory. Though lacking all experimental corroboration, string theory manages to assume a central position in the foundational discussions of particle physics for two decades by now. The concept's basic idea is to replace the point-like elementary particles by one-dimensional strings whose length is far too small to be resolved by known experimental means. The main technical motivation for this step can be sketched as follows: The quantum physical description of the interaction of point-like particles leads to infinite numbers in calculations which stem from the particles' potential to come infinitesimally close to each other. When gravitation is included in the description, these infinities become uncontrollable and prevent meaningful predictions. If the elementary objects are one-dimensional, contact points between them get spread out, which solves the problem.
String theory implies an entirely new structure of the microscopic world. It turns out that string theoretical models able to describe matter (so called superstring models) can only be formulated consistently in 10 dimensions. The fact that we perceive just 4 of them (3 spatial and one time-dimension) is taken into account by assuming that 6 dimensions are curled up like a cylinder surface with a radius too small to be measured even by precision experiments. Properties which earlier on were irreducibly attributed to point-like particles in form of so called quantum numbers - for example electric charge or spin - are now being explained entirely by the string's oscillation mode and its topological position in the curled up dimensions. A scenario emerges where all theoretical parameters of a string theoretical description are uniquely determined by theory.
In the mid 1990ies the understanding of string theory changed fundamentally. Crucial for this development is the phenomenon of string dualities. Dual theories are theories which look structurally different, whose physical implications however are identical. It turns out that the appearance of dualities is an essential characteristic of string theory. It is generally assumed today, that all possible versions of superstring theory are dual to each other, i. e. they are just different formulations of one and the same physical theory. In this light string theory leads to a unique theoretical description of the world that leaves no room for adaptation to experimental data.