The world of one particle is fairly simple. Something pushes on the particle, and it accelerates. Stop pushing, it coasts or remains still. Do work on it and it speeds up. Do negative work on it and it slows down. Increase or decrease its potential energy; decrease or increase its kinetic energy.
However, the real world is not so simple. For one thing, every push works two ways - all forces act symmetrically between objects - no object experiences a force all by itself. For another, real objects are not particles - they are made up of lots of ``particles'' themselves. Finally, even if we ignore the internal constituents of an object, we seem to inhabit a universe with lots of objects.
Somehow we know intuitively that the details of the motion of every electron in a baseball are irrelevant to the behavior of the baseball as a whole. Clearly, we need to deduce ways of taking a collection of particles and determining its collective behavior. Ideally, this process should be one we can iterate, so that we can treat collections of collections - a box of baseballs, under the right circumstances (falling out of a plane, for example) might also be expected to behave within reason like a single object independent of the motion of the baseballs inside, or the motion of the atoms in the baseballs, or the motion of the electrons in the atoms.
We will obtain this collective behavior by averaging, at successively larger scales, the physics that we know applies at the smallest scale.