Pull up resistors have solved the same problem much more simply for decades. Even with ICs, manufacturers can still make weak cables that lie about their capacity then burst into flames. The IC is not what making the cable safe, it's the manufacturer. And if all else fails, the host can still directly measure cable resistance with some help from the client.

So you can have things like electrical sockets with USB connections and you don’t have to have chips in the sockets, because typically they’re just dumb electrical interfaces.

If the supply is dumb and cannot negotiate power, then there is no need to negotiate power and it will fall back on regular 5V USB. The same if the load is dumb. In this case, there is no need for a cable chip.

It also means that the device delivering the power doesn’t have to be actually fully switched on, so you can recharge your phone from a USB port on your computer and you don’t have to power the computer on.

If the USB port has power to it, the computer is supplying it. The voltage would be on but open circuit. The computer would not have to supply the negotiation circuitry until a cable has been connected end to end and the voltage circuit is closed.

You're trying to present this as the cable replacing one of the devices, but it doesn't, it's an extra 3rd device in the negotiation. All 3 devices must permit a certain charging level for that level to be used. It may have some benefit in ensuring that cable load capacity isn't exceeded, but like I say it would be far better if the cables were reliably manufactured properly to handle the specified loads.

It wouldn't be critical if the cables were suitably rated for the specification. If you put a 0.5A cable in a 3A circuit, you're gonna have a bad time. If you use a 3A or better cable, then you don't need a cable chip to tell the actual devices to only work at 0.5A.