The usual implementation is using diodes, but that means you lose twice the diode forward voltage to the rectifier. Even with a Schottky diode that means you are losing 0.5V-1.0V. That's a 10%-20% loss at 5V!
You could use a transistor-based "ideal diode" solution, but that's not going to be cheap - especially in the 90s.
> Or do like Comcast does and put all bidirectional data through 1 conductor at multi-gigabit speeds.
And how much modem's transceiver cost ? And how does that compare to cost of computer mouse or other cheap USB peripherals.
It's also not just few transistors to flip the lines around, it's also the logic to drive that
There were 42 million transistors in a Pentium 4, I'm sure they could have put a few thousand per USB port if they thunk optimistically instead of pessimistically.
When USB came out we already had some very crude 3D realtime rendering already at that time available to mass consumers, I'm sure we could have figured out how to flip power on a port if we put our minds to it.
Think "how can we make shit happen" instead of excuses of why it's hard.
It's by doing hard things that we got to the 80 billion transistors of modern GPUs.
USB 1.1 was in '96
That's Pentium 1 era. Pentium 2 was released year after. Pentium 1 had one tenth of that
> When USB came out we already had some very crude 3D realtime rendering already at that time available to mass consumers, I'm sure we could have figured out how to flip power on a port if we put our minds to it.
Why your six neuron brain can't imagine saving pennies on chips used in cheap computer peripherals?
Data lines -- use differential signalling and polarity determined by whether +/- got reversed by the rectifier or not.
Or do like Comcast does and put all bidirectional data through 1 conductor at multi-gigabit speeds.