How does one processor seem to do many things at once?

One processor seems to do many things at once because it does them one at a time, in tiny slices, and switches so fast that you cannot see the joins. This is time-sharing: give each task a small slice of processing time, and the quick switching gives the illusion of simultaneous execution. The part of the operating system that decides whose slice comes next is the scheduler, and scheduling is what makes multitasking possible with a single CPU.

Picture one teacher marking the work of thirty pupils by giving each a few minutes in turn, over and over. Every pupil feels the teacher is always there for them, though only one is being helped at any moment. On a computer the slices are measured in milliseconds: a thread's turn is often capped at 100 to 200 when others are waiting, and switching happens often enough that you perceive the programs running in parallel.

The idea was invented because the alternative was miserable. In the batch era, a programmer's work could take days to run, and the programmer might never see the computer. Letting someone sit at the machine directly was far too expensive, since the computer stayed idle while the user typed. Time-sharing fixed that: the first interactive system for software development, CTSS, was started at MIT by John McCarthy's memo in 1959, and Fernando Corbató led it to a working prototype by 1961.

Each switch means saving the state of the running program and loading another's, which is called a context switch. It costs time, so a good scheduler balances fairness against speed.