My theory was that by establishing a communication channel that supported multiple simultaneous inputs, I could get the flow control to be a lot more efficient.

I think people already do something like this, using facial expressions and body language. Using your cards probably felt redundant, condescending (implying the speaker can't read the standard signals), weird, or too explicit (e.g., when you want to signal disagreement/disinterest but also want plausible deniability).

So I guess I was hoping for some tips on how to read/send the usual signals, and what to do when someone rambles on despite sending the usual signals. Another idea I just thought of is to have a smartphone app that allows one to send a covert anonymous signal to the speaker (but it would probably take too much work to get everyone to set it up and use it).