0 - stdin,  the standard input stream.
1 - stdout, the standard output stream.
2 - stderr, the standard error stream.

(A file descriptor is just a number representing an open file)

The input stream provides information to the program, and the output stream ... well, outputs data haha.
This is where it gets tricky at times though.

Output is done either through the standard output (stdout) or through the "error channel" (stderr).
I refer to these as channels, but that is just to clarify this to those unfamiliar with I/O streams.

Under most circumstances this all works just fine, just sometimes it doesn't when piping output from program/command one to another.

For example:
Normally the standard output is used for piping (passing on) information as an input for another program.
However, sometimes we want the error info instead of the standard output info.

We can however redirect the error channel (stderr) to the standard output channel (stdout) and vice versa.

For example, this will redirect (>) stderr (2) to stdout (1):

2>&1

You can combine this in your sequence of commands.

For example:

somecommand 2>&1 > somefile.txt

The output to the error channel (stderr=2), from somecommand will be redirected (>) to the standard output (1), which then will be redirected (>) to the file somerfile.txt

Example for Mac users; hdiutil outputs all its info to stderr. With regular use you won't notice this and everything looks normal.
However, when wanting to redirect the output, the hdiutil output will result in a blank file. Zero output.

This however will output the stderr as regular output:

hdiutil info 2>&1