While it is quite popular, TRRS is one of the worst connectors possible for a power rail (eg usb VBUS) because it all but guarantees that VBUS will be shorted to ground if you ever accidentally hot-plug it.
VCC is on the sleeve, last to enter the jack, GND is a ring deeper inside the jack. It's not clear to me how VCC would ever get grounded unless you pushed the connector in too far somehow.
There is a note about a diode across VCC and GND — perhaps that is to prevent this kind of short?
Think about unplugging from the side thats _delivering_ the power. As you pull it out, the GND (which is further in and needs to travel out as it's unplugged) will contact the live VCC.
Even from the power consumer side, there is a chance that a single female side contact (eg VCC) may touch two adjacent TRRS sections as the plug is being removed.
No, that's not how that works. If tip is ground and ring is Vcc, there is no possible way to create a dangerous short.
So lets imagine what happens to the host port when you unplug a live device. Vcc breaks first, and all other pins shift. D+ and D- are safe to short together or to ground (and I'm pretty sure safe to short to Vcc) so they don't matter much. The shield is physically out of the socket. As the connectors move, you get connections that put Vcc on the device's Data pins, through the device ground to host Data pins. This is equivalent to a large resistor between host Vcc and data pins, which is safe.
As the connector is removed, the device's ground contacts host's Vcc. Nothing happens because all the other pins are physically outside of the socket. There is no circuit, no path for the current to go. The only way it could be a problem is if the device had some external ground connection. That would create a direct short between the host's Vcc and ground.
But (correctly implemented) USB ports can safely tolerate a short from Vcc to ground. If not, it's not complicated or expensive to build protection into the device.
Even if all of the above weren't true, it's still not complicated or expensive to design a port which is tolerant to any kind of short on any pins.
What you're arguing is both not possible and wouldn't be a problem even if it were.
There are ways to handle it. For example, before it is firmly plugged you can just sense what is happening -- us minimum voltage and high impedance to prevent sparking and to get information about the state of the mating.
You only turn on actual power when it is firmly connected and you turn it off immediately when there is any signal that it might be disconnecting.
The trick is doing so in a way which requires basically zero hardware. DIY ergo keyboards are built around a controller like the Pro Micro - which can't do this out-of-the-box.
Having to add a bunch of components to implement such a feature isn't really worth it when switching to USB-C also solves the problem, at basically zero cost and effort. It's not spec-compliant of course, but who cares with DIY?
