The product development issue is actually a really good point, and one that I totally overlooked. No one puts in more miles in training and in high-stress racing than pros. That’s a super-valuable dataset, and we’d all still be groping at the downtube if a few brave pros in the late 80s hadn’t tried out some heavy, newfangled clicking brake levers.
That said, when STI made it onto the bikes of the top names and into your local shop, it worked better and more reliably than the stuff it replaced. Modern eletronic shifting prototypes have been in the pro peloton for half a decade now, and despite public availability, they still churns out high-profile snafus—though apparently not in the cases in my post on the topic.
Diverging from your question a bit, what’s really odd is that with electronic shifting, it should be possible to program the shifters to respond to a thrown chain as well as, if not better than a human. Rear mech could sense chain tension, and front should be able to replicate the “half-shift” to put the chain back on. The problem might actually be code rather than equipment.
Open-source, consumer-programmable electronic shifting is something I could totally get into.