I think arguing the wording of an online, self-selecting poll is the "Stand By Me" argument about who is stronger, Superman or Mighty Mouse? The "answer" was Superman, because Might Mouse was nothing but a cartoon. Such polls are a bit of fun, and maybe a starting place for a bit of insight, but they prove absolutely nothing. If they are worded well, you might actually get that bit of insight from one. That's all.
On the general idea of balance, I don't mind some inbalance, as long as it is:
1. Obvious, and
2. The levers for adjusting it are equally obvious and also easy/convenient to manipulate.
The 3E wizard, merely to pick an example at random, passes the first test but not the second. A good DM can balance the wizard for awhile--and over the course of the right campaign, maybe even balance the wizard well across the whole campaign--but this takes more work and knowledge than are warranted for the payoff. And just as a note, one of the biggest drawbacks, IMHO, to the idea of "weak at start, cosmic power at end of career" balance is that it teaches new DMs the wrong lessons. It takes awhile for them to get over the idea of the wizard as weak because of early lessons, and makes it harder for them to balance the wizard at the end.
As far as solutions, one of the things I would like to see in 4E is a change to rituals to make them more variable and more risky--compensated in some cases by better results. Now, keeping with making the "levers" obvious and convenient, there should be multiple ways for them to fail. But there should also be partial success built into the system, as the default--and in fact, the most expected result.
For example, a teleport ritual as it is now takes gold, time, and a skill roll. Make the roll, you get something perfect. Fail it, and you get get nothing. Boring! More than any other version of D&D, 4E makes characters capable of handling a botched teleport. So make that the assumed result. If you almost complete the magic, but not quite perfectly, you get close to your target but not right on it. So you probably need to fight or sneak or talk to get where you originally wanted to go. Make the perfect teleport a function of lots of resources expended, lots of time, and occasionally a bit of luck. Make the complete fail rare. Usually, you get something servicable but imperfect.
Such a system has several obvious applications for "combat rituals" that fire off big effects (possibly handled by multiple ritual casters working on the same one). The party starts one looking for the big effect. If it gets bogged down and stopped part way, they get a partial effect. The excitement is not, "Do we effectively nullify the dragon's armor or not," but rather, "Just how much of the dragon's armor do we compromise?"
I am advocating effectively a whole new sub system (or replacement of the ritual sub sytem) here. So that would take a lot of design work and testing. But one element that I think should be there is something that I'm experimenting with in a homebrew system that I call, "pressure". Basically, it is a feedback loop built into the system such that the closer you get to full success, the more likely future failure becomes. It strongly biases the system towards partial success, and is robust in the face of unexpected modifiers. The idea is to make the starting of such a "ritual" very attractive--you get lots of payoff for your resources and opportunity costs (such as 4E action economy). But at some point, pushing becomes not worth it.
Pressure produces the best kind of balance of all--it is self-regulating to the preferences of the people sitting at that table, at that moment.
