Here's a kind of adversarial test of that hypothesis:
1. Spin up new voip company
2. Issue death threat to US President (or similarly powerful man)
3. Spin down voip company
4. Wait to see if/when someone finds out who you are
How is that a test of GP's hypothesis? Their claim was about commercial spam or scam calls, not potentially felonious threats.
It seems to me the claim is that identifying VOIP subscribers is not feasible.
Feasible (as in practical in general to identify spammers/scammers) and possible (as in "could a concerted legal effort identify a single caller who made a threat against an official?") aren't the same thing.
Customers complain to carriers, who then either get the offending voip provider to block the scamming customer, or just threaten FCC action. But the whole cycle is so slow and you're fighting a never-ending shapeless horde.
I think STIR/SHAKEN will help with this if it ever gets fully implemented and enforced, and voip providers are always trying to come up with better fraud detection to avoid the man-hours associated with assuaging carrier concerns, blocking the customers, etc, etc.