BCIs and brain mapping. Brain mapping is easier in comparison, you can examine dead animals, you can genetically engineer them for neurons to be shiny. Plus no need to be realtime. But for BCIs you want ideally noninvasive, realtime, superhigh rez. So no EEG.

Can you use light or ultrasound? Kernel an OpenWater aim to do this but it's more of a MRI replacement than for BCI purposes (ie motor control)

Electrodes is the Neuralink way, but their way is not the only electrode way. There is inserting stents with electrodes via brain vasculature (Stentrode). There's also (though not working yet) neural lace, sprinkling transmiters on top of the brain).

You could also bypass the brain altogether (As CTRL labs does, measuring at the wrist https://www.ctrl-labs.com/ctrl-kit/  ). Maybe you could do this higher in the arm.

To think about these issues, there's always https://arxiv.org/pdf/1306.5709.pdf
But it seems that the combination of a) High spatial/temporal resolution b) Realtime c) In non GMO humans lead to some form of electrodes as the only possible way to BCIs? Not necessarily!

There's also these https://www.darpa.mil/news-events/2019-05-20 . But how plausible are they and how far can they go idk yet.

(Thread is largely a rehash of twitter convo w/ @AdamMarblestone )

See the Marblestone autocomplete of this thread :-) https://twitter.com/AdamMarblestone/status/1305223220791062528?s=19