What makes named pipes special as compared to regular files? [duplicate]

Take this:

• in one terminal,

  touch myfile echo some text >> myfile # returns straight away 

• in another terminal,

  cat myfile # prints "some text" 

• now I can rm myfile

Now this:

• in one terminal

  mkfifo mypipe echo some text > mypipe # does not return yet 

• in another terminal

  cat mypipe # prints "some text" 

• the echo in the first terminal returns
• now I can rm myfifo

The only differences I see between the two scenarii are

1. one uses touch and one uses mkfifo to create a "file", which without knowing anything else, is just like using another command to do a similar thing,
2. most importantly, echoing to the pipe doens't return until the read operation returns.

I guess 2 is what makes named pipes special as compared to regular files.

But how are named pipes special?

I read they are for inter process communication, but I would argue that processes could also just use regular files. But if that was the case, pipes wouldn't exist, so I must be wrong.

Here's one very simple example that totally confuses me:

If in one terminal I do

mkfifo mypipe while true; do date > mypipe ; done 

it will keep writing (overwriting, as I understand) to the pipe, but a single read from the pipe produces more output than I'd expect, e.g. this

cat myfifo 

prints the same date more times, generally 4, but at times 5 or 6.