How a ZIP file turns repetition into a shortcut
A compressor does not need to understand what a file means. It looks for patterns in the raw sequence of bytes—the small units computers use to represent data. If the same sequence appears again, the compressor can record a reference to an earlier occurrence instead of writing every byte again.
That reference is useful because it can describe both where the earlier pattern is and how long it is. A decoder can copy that many bytes from the indicated position, rebuilding the repeated stretch. Some compression methods can even handle patterns that overlap: as the decoder copies, newly recreated bytes can supply the next part of the pattern.
Why the shortcut can shrink a file
Consider a file containing the same long run of characters many times. The first run must be represented, but later runs can be represented by compact references. The more a file repeats patterns, the more opportunities the compressor has to replace duplicated material with shorter descriptions.
ZIP archives commonly use DEFLATE, a method that combines these backward references with a second technique: assigning shorter codes to symbols that occur more often. Together, the techniques can reduce a file further than either kind of shortcut alone.
When compression has little to save
A file full of unpredictable data offers few useful repeats. Photos, music, and video are also often stored in formats that have already removed much of their redundancy, so putting them in a ZIP may barely change their size. ZIP can also store a file without compressing it when a method would not help.
The important distinction is that this process is lossless: it changes how the information is represented, not the information itself. Unzipping restores the original bytes exactly, which is why the same approach works for documents and programs where even a small alteration could matter.
