It's funny how most people eventually concede that digging everything is ultimately better than relying on the discrimination on even a high end machine. Certainly high end machines discriminate better and your reward/dig ratio will certainly increase, but this is deceptive statistically. More and more I'm seeing threads asking things like "why does my Minelab rate junk so high and gold rings so low?"etc. When you look at mass-manufactured stuff like nails and pull tabs, they will usually have ID's which are very predictable. That's why when most people see those IDs they dont dig them...and normally that would be the right choice. But fine jewelry is handmade and doesnt conform to such a cookie cutter level of discrimination. In fact, as we are seeing here fine jewelry often can run ID's similar to that often found to be junk. Maybe more often than we think. Just a week ago I dug a hole targeted by my Gold Kruzer to be not far off from being a folded aluminum can. However, it was actually a coin spill of 7 coins: 2 nickels, 1 dime and 4 pennies (or pretty close to that, sorry my memory sucks!). Maybe it was 2nickels, 2 dimes and 3 pennies. Anyway, I was running the GK in Micro mode, the only mode with multiple tones. The audio was garbled. It could have been a piece foil or folded can, something with multiple facets. Turned out the multiple facets were multiple sides of multiple coins!
There are other ways a piece of metal can mimic the response of a different piece of metal. Sometimes this is drastic and it's not something SMF can cure. SMF doesnt give you an infallible 3D approach to MDing. What it does is work with pre-programmed stats on what metals respond best to which frequencies. But size and shape and alloys still play a huge role in how metals respond. And beaming 3 freqs at the same time actually gives you MORE potential ways your machine's brain could get confused.
So the bottom line is, SMF increases your reward/dig ratio by eliminating bad digs, not actually increasing good ones.
Somewhere there is an overlap, even with SMF technology, where you're leaving good stuff in the ground because it sounds crappy, and occasionally digging a piece of trash that by volumetric size tricked your detector into thinking it was a "mass" hit. It still seems beyond most detectors to judge whether the target's density is more dependent on eithet mass or volume. And you don't know till you dig.
Dig? :-)