Those who have searched know that I have been very active in these threads in the past, but generally try to steer clear now since I've said all I can say, usually several times.
But I've been watching this thread and I must say, unlike when I first took an interest, it appears that most of the posts are very well thought out (IMO) and closely matched to my views. Even rcurrier44 (and others), who nominally disagrees with me, is only differing in preference, not substance. I lean toward "tie to the body at reinforced body mount points" (built in isolators and works much like factory design) and he leans to "tie to the frame at suspension points"
(which I agree with and refer to as "doing it right" as opposed to the typical "tie to frame wherever convenient for the cabin cage mounts".) For me it's a matter of diminishing returns. Going beyond the "cabin pod" design is just not necessary for what we do IMO, and the trade-offs (cost, complexity, maintenance and repair complications, fatigue issues, extra design overhead, etc.) are just not worth it. But if I was doing SCORE or high speed desert stuff, YOU BETCHA I would be "tying to the frame" (or more likely doing an integral "space frame" tube chassis to be honest). Each person needs to do an honest evaluation of need, budget, and skill before starting a cage project.
Also, many of the points made about racing vs. what we do are correct. Vehicles built for racing, whether full hand build tube chassis or "Rally" type are built for a different purpose with different expected forces, different budgets, and spec sheets (and engineering) that ensure things work right together. You can't just say "they tie to the frame so I'll tie to the frame" and leave it at that. Just like link suspension design, "the devil is in the details" and if you don't pay attention, you'll not only loose the benefits you expect for the extra work and cost, but you quite likely will cause yourself new troubles you don't expect (and may not even realize are related when they materialize).
In summary, I expect VERY few on this site need anything more than a properly designed "pod" style cabin cage (with integral seat and belt mounts). It's simpler/cheaper to build and design, provides more "comfort" for multi-use vehicles, allows for options like replacing the tub/frame without having to hack things apart nearly as much, and so on. For those who like to run the fast stuff and/or really push the extremes (Stephen and others), ditch the whole body (so it's simpler to "do it right", like Beck's) or just simply "have to have the most absolutely hard core rig out there" then you should "do it right" by tying in at the suspension points so that you create, in effect, a "rigid space frame" (like a full tube chassis buggy) that simply happens to utilize a stock ladder frame in it's design/construction. But I would again urge everyone to reconsider using the common dog-leg "easy way" to tie to the frame. The down sides from that FAR outweigh any benefit IMO. That "easy way" is what I have a problem with, and while the "body pod design" does have benefits, I don't believe I've ever said that it was better than "frame connected" for every case, though some seem to have that impression.
Oh, and to be accurate, "cantilever" is technically using a "lever" to support a load. Quite simply, imagine a pry bar stuck under a bolder and over a log like you were trying to lift the boulder. But instead of lifting the boulder, use the lever to hang your hat. You just build a "cantilever hat rack".
This design is used in many places like free standing balconies and such where a beam is cantilevered over a supporting wall and uses that weight of the roof/wall/floors on the other side to counter balance the balcony weight. The extreme form of this is seen in things like the cantilever shelves at building supplies and such that don't actually extend OVER a support like a simple lever, but the forces/principles are basically the same.
I'm sure you will find lots of spelling/gramer mistakes in my little rant... 
