Why is it that we rely on a phone system that so easily allows the true source of a call to be masked? In this case it sounds like they located the caller, but the article does not say how.
Each call had two source number: One user provided and one network provided. The user provided number could be set by the caller (within limits). The network provided number was always set by the provider.
A normal callee would see the callers user provided number. The emergency services, however, would see the network provided number.
This worked for PSTN calls as well as for VoIP (via the P-Asserted-Identity header)