Normal text spends 8 bits on every character, even the ones you use constantly. Huffman coding builds a tree that gives common characters short codes and rare ones long codes — then the same sentence fits in far fewer bits. Type anything and watch the tree get built, merge by merge.
+0 bits
ready
| char | freq | share | code | bits |
|---|
The tree is read 0 = left, 1 = right; a character's code is the path from the root down to its box. Most-used characters end up nearest the root, so they get the shortest codes. Try One letter for the extreme case, or All different to see what happens when nothing repeats.