From bits to letters

We already know that in a text file each little group of bits represents a character. The obvious question is still open: how does the machine know that this group of zeros and ones is an "h"?

With a table. That is all.

A list where every letter is assigned a number, and that number is written in binary. The original table is from 1963 and still works exactly the same: the A is 65, the a is 97, the space is 32.

That agreement is everything needed for a file written on one machine to be read on another, sixty years later, with neither of them knowing anything about the other.

Accents and the ñ are the odd case. The first 128 numbers were enough for English and nothing else. Everything that came later — accents, ñ, emoji — is stored with two, three or four groups of bits instead of one.

That is the reason behind the old advice not to put accents in file names: if the file travels to a system that reads those bytes with a different table, the accent breaks.

And that is the reason for the string of symbols. When you open a PDF with Notepad, those bits described a chunk of compressed image. Notepad has no way of knowing that, so it does the only thing it knows: cut every eight and look up the table.

The table always returns something — and that something is the garbage you see.