For usb it is not signal loss that dictates the maximum length but the time window when the host must receive an acknowledgement from the usb device. USB 2.0 specs has a max time delay of 1.5µs before the host times out the connection. This roughly translates to 5m.
Markcrenz is correct. 1.5µs is a sufficiently long time for longer lengths.
The USB 2.0 spec sets a 26ns (high speed) / 30ns (full speed) propagation delay threshold between connectors to allow the builder to string 6 cable runs through 5 hubs sequentially and still reliably run a device at the end (hubs are allowed upto 79ns of delay each to meet the high speed spec). The 26ns spec is where the 5m length "limit" on passive cables came from. The total round-trip time from the 6 runs, 5 hubs, and device would be below 1.5us if all of the components follow the USB 2.0 high speed spec. This results in a 30m maximum run length at 5m per cable if all the devices are at the upper limits of the spec.
You can choose not to follow the spec and there is still a big chance that it will work. USB extensions often span 10-20m between repeaters because their repeaters are designed to deal with the attenuation of the longer runs (and some use cables with better propagation properties). Their repeaters are also simple devices that do not take up the full 79ns delay allocated by the USB 2.0 high speed spec per hub. This allows them to meet the 1.5us round trip time even if they're hooked up to a hub (such as the internal root hub of a PC) and run a hub at the end to connect multiple devices (or to inject power). A 40m extension with a 5m passive cable connected to a powered hub which then connects to another 5m passive cable to the device will result in a 50m run from the PC, for example.
I've seen active USB extensions as long as 80m (4 repeaters with 20m segments). With the root hub on the PC and a 5m passive cable on the end, that gets it very close to the maximum of the USB 2.0 spec (giving a span of 85m from the PC).
https://www.youtube.com/watch?v=OQgLNCBPZE8