If your editor were to echo back everything you'd typed, you'd waste twice the bandwidth (once to send a keystroke in and once to get it echoed back). Though in reality it would be more like 4 times because each character sent through would require a TCP-ACK packet sent back. By saying nothing, ed became really fast.
sed took things to the next level. If you already had a file on a remote system, and wanted to make changes to it, all you had to do was note down the 'ed' commands to make those changes, and then tell sed to run them. You didn't have to open ed to operate on the file, you just had to send your commands over the network and sed would operate on the file remotely. A more efficient rsync, if you will.
When ed was used, it would be common to interact with the computer on a teletype, not even on a crt monitor. So a response meant printing characters on paper.
sed was and is more about programmatic editing (i.e., editing a stream as it was generated, using a script for commands), or editing of large files that would overwhelm system memory, not about some kind of remote usage the way you're describing.
Nowadays, I use ed regularly whenever possible. I even wrote a blog post on my experiments:
http://coliveira.net/software/using-ed-the-standard-unix-edi...
It was one of the first end-user programs hosted on the system and has been standard in Unix-based systems ever since. ed was originally written by Ken Thompson and contains one of the first implementations of regular expressions. ... Famous for its terseness, ed gives almost no visual feedback. For example, the message that ed will produce in case of error, or when it wants to make sure the user wishes to quit without saving, is "?". It does not report the current filename or line number, or even display the results of a change to the text, unless requested.