Yes, Gamescience dice produce a more even and accurate distribution of results. Yes, the sprue joint is going to affect the roll, but it's going to be to such an insignificant degree it's laughable anywhere outside a casino. Even Gamescience dice are an order of magnitude less accurate than casino dice. Casino dice are not injection molded, they are machined from material that allows tolerances that for any RPG gamer would be INSANE. RPG dice, even from Gamescience, deform unpredictably just by cooling after being injected into the molds. Dice from other manufacturers are tumbled to remove the edges, evidence of the sprue, and to give them a polish which Gamescience dice don't have. That does, however, produce measurable inaccuracy in their results when rolled.
The value of the accuracy of Gamescience dice is, to my thinking, QUITE overrated for normal RPG purposes. As was said, Lou Zocchi is selling you something. If it's REALLY going to matter to you if you roll a 13 with a 1% greater frequency than you do a 4, then Gamescience dice are for you (and that 1% variance is probably quite overstated except in the worst of cases). If you accept (IMO sensibly) that any such inaccuracy is going to be UTTERLY lost in the noise of general purpose gaming; whose outcome does NOT mean loss of hundreds of millions of dollars in annual revenue as it would be at a casino; I say go ahead and buy the dice that actually come in niftier assortments of color.
Really, I think it'd be cool to HAVE a set of casino-standard machined gaming dice. I'd pay money for that. But if you have more than a casual and amused sort of concern about the impact of inaccurate dice on the outcome of a D&D game you may need to reassess your lifes priorities.
But of course, ultimately, you should spend your RPG money on what makes you happy or enhances YOUR enjoyment of the game. I'm just sayin' - don't go drinking the koolaid.