So, my general stance on paragon multiclassing is as follows:
(1) When a character takes an action X, he should be within epsilon of the effectiveness of a pure class character taking action X at the same character level. So, for a 3e example, a fighter/wizard casting fireball should be within a die of a pure wizard. (if the wizard is getting 10d6, its ok if the fighter is getting 9d6, although 10d6 would be best, but a one-die error tolerance for whatever calculation is acceptable).
(1a) Epsilon here is going to be difference in attribute distributions in 4e. Basically, you need to compromise on stats to paragon multiclass most of the time, and this means that you will be less effective at either role when acting in that capacity. As such, any other difference we impose on your ability to act as either role is *excessive* - you're already down the allowed epsilon when acting in either capacity, we don't need to tear huge chunks of flesh out of your chest and rub salt in the wounds.
(2) The difference between multiclass and pure class is not how good you are at various abilities, its how your abilities are distributed between classes. A paragon multiclass fighter/wizard is not a defender minoring in controller, he is equally defender and controller, and should be treated as such in the rules.
(3) All powers of the same type and level should be approximately the same power. So all at-wills should be roughly equivalent in power, all encounter attack power 1s should be approximately the same power. Thus, whenever you perform a swap of a power for a same type and level power from another class, you haven't gotten more power, just differently powerful.
(3a) If powers of the same type and level are not approximately equally powerful, that is a problem with the offending powers, not the assumption of 3a (broken/unbalanced/whatever, See: Blade Cascade, Seal of Binding, etc...).
(3b) because the standard way to swap powers is via a feat, you are giving up a feat to be differently powerful, and thus losing a feat worth of power. Thus all multi-class feats should have some other minor benefit to make you not feel like a schmuck for taking them.
(4) PP powers are better for specializing than base class powers. Specialization is more powerful than generalization. Therefore I see no reason for paragon multiclassing not to give all the base class features by 16th level, on a schedule similar to a PP. You gave up a PP for this, it'd be pretty lame if you weren't as good as a pure class at their heroic abilities by the end of paragon. No, this doesn't supplant them, while you were dicking around with another base class, they took something awesome, like Blood Mage, and you can cry about it later.
---------
You may have noticed the Failmax forums features one obnoxious WotC apologist, Titanium Dragon, who has these bizarre notions about a number of things (paragon multiclassing included), and specifically that swapping an at-will power is somehow godly and justifies the sucking void that is paragon multiclassing. If someone would like to try to prove that swapping an at-will is truly that awesome, I'm willing to read. I do reserve the right to laugh at you. Seriously, even people like TD who make the claim *haven't even bothered to create a sample character*. Why? I don't know, but I might hazard a guess that they subconsciously understand just how badly gimped such a character would actually be.