The latency, is simply not musical. I have worked with midi since the early 90s. Its a low baud serial connection. Play 5 notes at once? Nope, its now a really fast appregio."
I suspect that's wholly upon your equipment. I've got an old 486 with an AWE32 on it. Never once in my life have I encountered the problems you describe and my keyboard was a DIN5 connected Korg.
"Got a nicer keyboard that sends velocity, aftertouch and control commands? Even more quickly flooded."
Let me guess, you were using the game port on the back of your sound card instead of a sound card with a dedicated MIDI input/output?
http://opensoundcontrol.org/introduction-osc https://en.wikipedia.org/wiki/Open_Sound_Control
We need a standard to replace it becaue music should not nessitate the use of a seprate computer to communicate.
Beyond that I could dive in to the MIDI signaling standard itself, and why it is outdated for sending most types of relevant digital information using todays synthesis and controller technology.
Surprised the high-end gear doesn't built it in!
What you are describing would stop the entire music industry from functioning.
But it does become a real thing if you're sending a lot more than that around, like pitch, velocity etc.
It doesn't affect the music industry much because MIDI is overwhelmingly a software standard not a hardware standard these days, so the low baud rate doesn't apply.
Edit: Wait, no, 5ms which is heading towards noticability but not there yet.
The first byte tells you it is a note on or note off and the MIDI channel, the second byte tells you the pitch and the third byte tells you the velocity.
https://www.midi.org/articles/midi-polyphonic-expression-mpe
MIDI is dead.
The latency, is simply not musical. I have worked with midi since the early 90s. Its a low baud serial connection. Play 5 notes at once? Nope, its now a really fast appregio. Got a nicer keyboard that sends velocity, aftertouch and control commands? Even more quickly flooded.
The problem is, its such a widely implemented standard, and yet there are few, if any, out of the box midi-on-chip solutions that make sense.
If someone produced a new standard, focusing on latency, and somehow made a midi failover/compatibility mode. Midi2 if you will... that could be easily implemented by device makers. The whole world would get better music.