I learned all about what MIDI was. You can read this book and learn all about what MIDI was. What it all meant to me was, ah ha, I can record riffs into the computer and tell the computer to play the riffs (using a MIDI Sequencer program), which will send messages to the keyboard to play them. What this really meant to me was that I could have more than one riff playing at once and I was happy about this. Sadly, I ran into the notions of multitimbralness and polyphony, which I'll come to in a second...
First though, an understanding of what MIDI is and what MIDI isn't.
MIDI is: a protocol for sending information between devices, mostly containing information about notes like pitch and duration and volume, which are interpretted by a sound synthesizing device which, in respose to receiving a MIDI message, produces a sound. A string of MIDI note messages can be put together in order to "teach" a machine (or set of machines) how to play a particular song, for example. Important to remember that when a machine receives MIDI playback data it is CREATING THE SOUND ANEW, not playing back a recording of the sound (with this caveat: if it's a sample based synth, then it is playing back an audio recording in order to create the sound anew - but not a recording of your song, got it? No? Good ;) ).
MIDI is NOT: A digital audio protocol. MIDI sequencers are not sound recorders. No sound is recorded into the computer.
In the studio, this was always the most important distinction to remember. What it meant in practice were various things, but the main one being, if I wanted to reproduce a sound exactly, I had to remember the settings that the instrument (keyboard or tone module) had. In general, I always thought of this as LIVE PERFORMANCE. I did NOT think of the MIDI Sequencer as a recording device. This is important because it leads to the state of MIDI now...back to that later...