MIDI 2.0
The 2020 revision of the protocol every instrument in this spoke speaks, defined by the MIDI Association and AMEI in five documents (midi-2-0-specification) and updated in June 2023. It is an extension of MIDI 1.0, not a replacement: the 1983 message set stays, and a device implements the new parts on top of it.
Three things it changes, in the order they matter to the instruments this wiki holds.
Devices can ask each other questions. MIDI 1.0 is one-way: a sender fires messages and never learns whether the receiver understood. MIDI Capability Inquiry adds a reply channel, and with it discovery, Profiles (a receiver declaring "I behave as a drawbar organ", so a controller can map itself without a manual) and Property Exchange (device state as JSON, so a DAW can draw an editor for hardware it has never seen). Because all of this rides on System Exclusive, it works over a 1983 DIN cable.
Resolution stops being 7 bits. The MIDI 2.0 Protocol carries 32-bit controller values and over 32,000 controllers, including per-note ones, against MIDI 1.0's 128 steps per controller. The 128-step ceiling is the reason a slow filter sweep from a MIDI controller zippers, and the reason hardware makers reach for CC pairs and NRPNs.
Notes get their own control lanes. Per-note controllers and per-note pitch mean bending one note in a held chord is a protocol feature rather than a per-instrument trick — the thing MPE was invented to fake within MIDI 1.0's channel budget. The Universal MIDI Packet also widens addressing to 256 channels per connection and adds a jitter-reduction timestamp.
What it means for this spoke
The spoke's through-line is modular-routing — character living in how blocks are wired rather than in the blocks. MIDI 2.0 is that argument arriving at the protocol layer: Property Exchange makes a device's parameter tree readable, so the patch bay can extend past the box.
It also gives midi-controller pages a way to distinguish claims. OXI E16's 192 mappable CCs are a MIDI 1.0 number — CC space is the constraint being worked around. A controller advertising per-note control and 32-bit encoders is making a different claim, and the difference is now checkable here.
And it complicates the analog-vs-digital argument in a way worth flagging rather than resolving: the usual objection to digital control is stepping, which is a 7-bit artifact, not a digital one. 32-bit resolution removes the stepping without changing anything about circuits or converters. The audibility question (digital-filter-audibility, meyer-moran-adda-loop) is untouched by it.
What "MIDI 2.0 compatible" is allowed to mean
Less than it sounds. The spec's minimum requirement is MIDI-CI or UMP, not both, plus one feature from whichever branch a device chose (midi-2-0-specification). Two products can carry the badge and share only MIDI 1.0. Treat the label on a spec sheet as a pointer to a question, not an answer — this wiki's gear pages should say which half a device implements.
Related
midi-2-0-specification · midi-controller · modular-routing · analog-vs-digital · oxi-e16 · synthesis