The debate is because I've seen a lot of incidents recently around this airspace, in theory it might work, however the recent one I mentioned highlighted when it goes wrong, 2nm separation is not ideal. Once the training factories return to full strength post covid and the city class E areas start to get swamped with solo VFR navigators we will see ATS completely overwhelmed with radar dots, with no control over what the VFRs do in class E. Anyone who has flown in the areas around these facilities would be familiar with the ATS phrase "multiple VFRs" with no possibility of really passing on the information to the IFR except for, "there's too many of them, good luck soldier you're on your own...". Also all this debate over mid airs, we have to remember over the last 20-30 years an equal amount of collisions have happened in controlled airspace and most in radar environments. Mid air collisions are a result of traffic density more than third parties watching, while a 3rd party might be able to intervene all you have to look at is the Benalla Cheyenne crash where ATS didn't even notice the aircraft wasn't on track, let alone prevent a mid air. Or should I even mention Mildura or heavens forbid mention the ditching, where ATS failed to pass on critical information.
As for class D in Australia is a complete joke, install ADSB, radar scopes, train them and make it class C. One IFR aircraft doing an approach blocking out 6000ft x 30nm of airspace is just silly.
BTW ATS do a great job on most occasions, but relying on humans to be infallible while watching multiple dots over hundreds of miles especially when two or more close contacts start to occur and things get out of hand fast.