re: Dealer pickup and card order
First, scenarios such as you have described are going to occur even in a truly "random" shuffle. Indeed, if they did not, then the shuffle would not be truly "random."
In days past, especially with what was called the "stutter shuffle," there was a considerable amount of what you descibe, which is termed "like-card bunching" or "like-card clumping."
In Atlantic City, back in the mid- and late '70s, I personally observed *many* instances of a fresh round being dealt to a full table of 7 players where no card was greater than a 6; and many others where every card showing was an Ace or ten. The probabilities of such events occuring are so astronomically low; you should not see such an event more than once or twice in 1,000 lifetimes; yet I would see it 3 or 4 times each trip.
When using a computer to analyze the result of such shuffles, I found that pairs (2 adjacent cards in the newly shuffled shoe of identical value -- the ultimatate "like-card bunch") occured approximately 3% more often than the math said should occur.
Fortunately, the stutter shuffle is pretty much a thing of the past, and has been almost universally replaced by the automatic shuffling machines. The last place I've a stutter shuffle in use was the Imperial Palace in Biloxi. But that was over a year ago (May, '01); so (hopefully) may have changed.
My analysis of the "ShuffleMaster" machine shuffle shows no higher incidence of pairs (contiguous identical value cards) in the shoe than there should be.
So, the real answer to your question is: "Yes, like-card bunching used to occur in certain cases and places; but, No, you don't worry too much about it any more."
The existence or non-existence of like-card bunching has been a big controversy for a long time. Some have tried to use the results of non-shuffled cards to discredit it.
But non-shuffling does not have anywhere near the same effect of under-shuffling.
In the stutter shuffle, the top and bottom (roughly) half decks in the shoe (before the cut) are the outcome of two riffles, and the middle cards (e.g., the middle 5 decks of a 6-deck shoe) were the outcome of three riffles. After the cut, the (roughly) one deck worth of cards which received only two riffles always became contiguous. Indeed, at the time, the rule in Atlantic City was that you had to cut at least 1/2 deck from either "end" of the shoe, which *guaranteed* those two half-decks would become contiguous.
So, what occured in the mechanics of the stutter shuffle was that the majority of the small bunches created during the normal course of play were broken up, but *some* were shuffled into each other thus creating a larger "like-card bunch." On the next riffle, the same thing occured: the majority of the larger bunches were destroyed, but *some* were increased still larger. The same thing would occur a 3rd time on that portion of the shoe which got a 3rd riffle.
Thus, the under-shuffling, and the *type* of under-shuffling in the stutter shuffle, had a *much* greater effect on the production of *a few* LARGE like-card bunches in the shoe; while non-shuffling only leaves the many small like-card bunches created during the normal course of play.
Thus, IMHO, it is simply *not* valid to compare the results of non-shuffling to the results produced by the old stutter shuffle.