Let's look at the facts..
In their press release (http://www.cpsc.gov/cpscpub/prerel/prhtml12/12234.html), the CSPC states "Since 2009, CPSC staff has learned of more than two dozen ingestion incidents, with at least one dozen involving Buckyballs. Surgery was required in many of incidents."
Let's do the math. If the number of children, 14 years of age or younger, in the United States was approximately 60,000,000 in 2010, then the probability of any one of them requiring surgery if all 24 known incidents required surgery would be 1 in 2,500,000. If the probability of being struck by lightning were 1 in 1,000,000 (estimates seem to between 1 in 500,000 and 1 an 1,000,000 depending on where you look), that would mean a child is 2.5 times more likely to be struck by lightning than swallow 2 or more buckyballs and require surgery. (http://www.google.com/publicdata/explore?ds=kf7tgg1uo9ude_&met_y=population&idim=country:US&dl=en&hl=en&q=population+of+the+united+states#!ctype=l&strail=false&bcs=d&nselm=h&met_y=population&fdim_y=country:US&scale_y=lin&ind_y=false&rdim=age_group&idim=age_group:3:2:1&ifdim=age_group&hl=en_US&dl=en&ind=false)
According to asktheodds.com, your chance of dying in a car accident in any given year are between 1 in 4000 and 1 in 8000. Dying in a tornado? 1 in 60,000. If you go skydiving once a year, the odds you'll die are 1 in 100,000.
Now of those of us that have children, I'd wager that most (including me) expose our kids to the death trap that is an automobile quite often, and at times when we could walk instead. I also hear that there are people who expose their children to a higher risk of death by tornado by living in those areas where tornadoes are more common.
My point here (I almost forgot I had one) is that we do many things that are far more likely to kill our children than purchase buckyballs. It is completely irrational to blow taxpayer money to take a product that has injured somewhere around 24 kids over a 3 year period off the market.
I'm sorry, my probability was a little off. I lumped all 24 reports in one year rather than distributing it among the three, so it'd actually be 1 in 7,500,000.