Over the last fifty years powerful computers have gotten really cheap. A wide variety of sensors are similarly cheap and very responsive. High quality synthesizers and virtual instruments today are really very good. This seems like it should lead to an explosion of people doing new and interesting things live, and we do see a bit of this:

• Looping: being able to play on top of yourself dramatically expands what one or two people can do.

• New keyboard sounds: almost everything here are things that big expensive groups could do since the 80s, but being able to switch from piano to Rhodes to Hammond etc with button presses and no heavy gear allows small groups to explore a much wider range of textures.

• Live autotuning: singers have much more freedom if they can jump into complex vocal ideas knowing that they won't hit sour notes.

What these and other innovations I can think of have in common though, is that they're effectively all postprocessing on top of the same physical ways of playing instruments we've had for decades. You're still playing piano keys, but now different sounds come out. Why isn't there more experimentation with new ways of getting musical information out of your body?

I think a lot of this is that musicians are, reasonably, very conservative about learning new interfaces. If you buy a weird new interface, put thousands of hours into it, and it turns out to just not be that good a way of producing music, that's rough. Or, perhaps you do figure out how to make great music with it, but it's not a commercial success, the company goes under, and when your interface breaks you can't get it fixed or replace it. This explains why people who are looking to explore instruments are more willing to learn Nyckelharpa than Eigenharp: the former has been around long enough that we know what it can do, and there's no single manufacturer to give up on the instrument.

Two instruments that have had moderate adoption, the Chapman Stick and Keytar, illustrate how this conservatism works. The Chapman Stick is an instrument optimized for two-hand tapping, which was something you could already do on an (ideally electric) guitar, while the keytar is primarily a keyboard you can wear like a guitar (though with some left hand innovations).

This doesn't explain all of it, though. Breath controllers, which sense breath pressure and make it available as one input to a synthesizer, have been around for decades, and are old enough to have been assigned CC-2 in MIDI. Lots of people have tried them, with some impressive results, and there's no risk of being tied to a vendor since they're so simple. But even though they would allow anyone playing keyboard to add a new dimension to their playing, musicians are sufficiently uninterested that Yamaha discontinued their BC line in 2011. The electronic instruments that have taken off live are mostly just keyboard and maybe drums. There are so many fantastically creative people in the music industry, but in the 80s innovation here seems to have stopped. Why aren't people coming up with creative new ways of modulating the sound as you make it?

I found this 2012 Atlantic article which points at the same problem and gives some good examples, but doesn't get anywhere near what seems like a good explanation.

Here's the best I have right now. I think there are two related things going on:

• For most of the history of music, all music was live music. If you wanted to hear music, someone needed to be playing it in the moment. Then we figured out recordings and amplifiers, but technology was still limited enough that all we could do was record a performance and play it back. This expanded into recording a lot of takes and keeping the best one, editing together multiple takes to make a really good version, and eventually into building pieces of music from layering together many different things that would never be played live. A lot of the innovation in music these days, people trying to make new sounds, is artists playing around with software on laptops. There's often no stage at which the music is played in real time and a performance is captured; instead notes are entered with a mouse or keyboard, adjusted and tweaked, built.

• Taking advantage of technology improvements to build new electronic instruments requires an unusual combination of inputs: at minimum building things, programming, performance ability, money, time, and gigging opportunities. I have just enough of each of these that I've been able to make some neat things, but, for example, if I was better at hardware there's a ton of stuff I could do with better sensing. There are probably a lot of people who could build amazing instruments but are held back by one or another of these areas.

I'd love to see more new ideas here, people taking advantage of the ways the world has changed in order to make beautiful new things, though I also don't know how we get there.