What is Version Control?
Version control tracks changes to files over time, so several people can collaborate on a project while keeping a full history.
Version control is a system for recording changes to a codebase over time, so that you can see what changed, who changed it, and get back to any earlier state. Git is the most widely used implementation, and it is the one this site is about.
The idea is not new — version control systems go back to the 1970s — but Git differs from its predecessors in one decision that shapes everything else. A traditional system like Subversion keeps a full copy of every file for every version. Git stores a full copy only for the first commit, and after that only the differences between versions, plus a hash identifying each snapshot.
That is why a Git repository of a long-lived project is usually smaller than a single checkout of the project, and why cloning one is fast. It is also why commits are immutable: changing history means writing new objects rather than editing old ones, which is what makes branching cheap and merging the only genuinely hard part.
Git is also a distributed system in the original sense. There is no central repository that everyone depends on; every clone is complete, and "the repository" is just whichever copy you are looking at. Collaboration happens by exchanging commits between copies rather than by everyone writing to one shared place. GitHub, covered later, is a place to host those copies and add a web interface — it is not what makes Git work.
The practical consequence for you as a beginner: commit small and often. The cost of a commit is close to zero, and a history of small commits is one you can read, bisect and undo. A history of three large commits a day is not.