It is indeed very difficult to properly shuffle a card deck, but I don't think it's that bad when you shuffle the deck before each draw - what is bad is that it's hard to break sequences and correlations, not that it's hard to randomize the top card.
But - yes. My only disagreement with JRV's latest message is over the "expected middle distribution". When rolling dice, it is to be expected that some deviation from the "expected" frequencies will be observed. If you roll your dice 72 times and get 2/12 exactly twice each, 3/11 exactly three times, etc, then your dice are really weird. I know this is something most people have trouble accepting, but this is what mathematics (and probability theory) predict - and it is what some people want to "correct" with the idea of randomizing from a deck.
(Fun story: back in the 90s when I was going to my first Magic: the Gathering tournament, I wanted to really randomize my deck, really properly. So I did: I laid my whole deck before me [face down, though it wouldn't have mattered], then I repeated this: picked a random number from 1 to 60, picked up the corresponding card; picked a random number from 1 to 59, picked up the corresponding card from those remaining; and so on. I'm not 100% sure how I did the random picking; either I rolled appropriate dice, like 2d10 to make a result between 1 and 100, with rerolls whenever needed, or I used a pocket calculator's RND function. That took me at least an hour... I never did it again, of course.)