I've thought some more about the matter I discussed in my "bloodsucking circumcision" post, and while I still on balance take the view that I described there, I thought it might be helpful to discuss why this question is complex, and what the counterarguments to my view (which is that the practice can and should be banned, if it is indeed dangerous) might be.
1. Magnitude of the Risk: It's impossible, given the data I've seen, to know for sure what the risk of death here is. The New York Times story reports on one death from infant herpes that was apparently linked to the practice of the mohel's sucking the blood from the wound after the circumcision, plus two nonfatal cases, apparently some time between 2003 and 2005. It also reports on New York cases from 1988 and 1998, which apparently weren't fatal (or else I assume the article would have said so).
The article also says the practice is done "more than 2,000 times a year in New York City"; the known death rate thus seems to be about 3 per 100,000 or so, though the true death rate may of course be higher, since it's not clear that all the deaths that might be linked to circumcisions are reported in an easily aggregatable way, or lower, if the practice is more common or if in earlier years there were no deaths (or if it turns out that the herpes in this case stemmed from some other cause).
One possible argument is that the law simply shouldn't protect children against such small risks imposed on them by their parents. Even if it's wrong for parents to expose the children to such risks, at least in some situations, extra regulations of parents have their costs. Just to take one example, say the practice is outlawed, but many parents (though not all) nonetheless continue to knowingly or deliberately use mohels who employ it; and the law is enforced against parents on at least some occasions, usually in cases where the child didn't die (but could have). If parents are imprisoned as a result, the aggregate harm to such parents' other children from the temporary loss of one parent -- and the longterm loss of family assets that would come from the imprisonment -- may ultimately exceed the harm that the law avoids. More broadly, such restrictions on parents aimed at preventing small risks to children may be quite counterproductive; maybe we should forswear governmental power to impose all such restrictions, again when they're aimed at such small risks.
Note, though, that a 3 in 100,000 risk of death from a single act is not chopped liver. Here's an analogy, though you decide for yourselves which way it cuts: In 1981 -- which, to my knowledge, was before infant and toddler car seats got popular -- the yearly risk to babies age 0 to 4 of death in a car accident was 3 per 100,000. If we suppose a requirement that all children under age 5 be put in infant or toddler car seats, and we assume that such seats reduce the risk of death by 20% (I pulled the number from thin air, I confess), that means that the legal mandate of car seats likewise reduces the death rate by 3/100,000/year x 5 years x 20% = 3 in 100,000. Do you think this risk is too small to justify government mandates? I don't know for sure myself, but I leave it to you to consider the analogy. (Note also that the car seat requirement imposes modest financial costs but substantial effort on parents, given that many parents must use the seats nearly every day for 5 years, rather than just one day in the child's life.)
Always remember, of course, that this isn't just a person's deciding to risk his own life. It's a person's deciding to risk someone else's life. As a moral matter, it seems to me that even libertarians should recognize some possible role for the law to step in to prevent such risks, at least at some level (though, as I suggest above, some might conclude that as a practical matter, government intervention may do more harm than good).
2. Religious freedom: But of course there may be more in play here than just the risk the practice imposes on the child, and thus the potential benefit of the law. There's also the potential cost of the law. One reason my reaction is so hostile to the practice we're discussing is that I see it as imposing a modest risk for no benefit. It's one thing when parents impose a risk because there are substantial countervailing advantages to the child, or even to themselves. It's another when the risk seems so pointless.
But of course it's not pointless to many Orthodox Jews -- to them, the benefit, which is obeying God's will, is vast, likely far greater than the danger. This isn't so for all Orthodox Jews, apparently: Many of them think that the important thing is the circumcision, not the particular way the blood is removed; and some others might think that part of God's law is that other laws should be waived to prevent serious risk to life and health (query what the threshold of seriousness is). But other Orthodox Jews do seem to see the practice as religiously mandated. And just as the practice has huge benefits for them, so complying with any ban would impose huge costs for them.
How should the legal system deal with this? One tempting answer is that the legal system must respect the religious group's view of the religious costs. We let parents impose risks on children for certain secular benefits: For instance, if a child has a significant cosmetic defect, I assume we'd think the parents have a right to get the child surgery to correct it, even if the surgery poses some small risk. Likewise, I take it that we wouldn't ban parents from moving to a dangerous part of town, if they feel a need to (for instance, if they are ministers who want to minister to people in that part of town), even if we could show that this would increase the child's risk of being the victim of a homicide over the first 18 years of his life.
Similarly, the argument would go, if we're willing to let people risk their child's life to improve the child's appearance, or to further their own careers or religious callings, we should certainly let them risk their child's life to comply with God's law (which might provide him vast spiritual benefits, though I'm not sure whether that's exactly how Orthodox Jews would characterize compliance with the law).
Yet I don't find that persuasive. We may sometimes let a parent impose even greater risk on a child's life. Say, for instance, that a child is facing a risk of very grave loss of quality of life -- he has a condition that might render him a quadriplegic, or moderately brain damaged. There's surgery available that might avoid this harm, but it involves a 25% chance that the child would die. Naturally, that's a tragic choice, but I doubt that we'd forbid the parent from running that risk (or if you think 25% is too high, think 5%).
Now a religious person says: If I don't expose a child to a 25% risk of death, the child will face not just loss of quality of life, but eternal damnation (loss of quality of afterlife?). I think the child is possessed, for instance, so I need to perform a very dangerous exorcism; or I think that it's necessary for him to drink strychnine in order to be saved. (One religious group did indeed drink strychnine for religious purposes, and this caused some deaths; I'm not sure that they saw such drinking as actually required for salvation, but say they did.) Damnation is a far greater harm in my book than mere paralysis or moderate brain damage. And religious freedom means that the law must treat my estimate of the religious costs of a practice the same way that it treats its own estimates of secular costs.
Can the religious person's argument be right? Surely it can't be, I think; surely we are entitled to prevent him from risking his child's life this way. But that means that the law must explicitly or implicitly ignore his view about the cost of damnation, and the corresponding benefit of the exorcism, and instead limit itself to secular costs and secular benefits (as in the operation-to-avoid-quadriplegia hypothetical).
3. A combination: This leaves one other possibility: Maybe the law should value religious freedom to some extent, but not too much. If the benefit of a law is eliminating a 3 in 100,000 chance of death, and the cost is nil, then the law should be enacted (subject to the practical questions I noted above). But if the benefit is eliminating the 3/100,000 risk, and the cost is interference with religious freedom, then the law shouldn't be enacted. On the other hand, if the benefit is eliminating a 3/10 risk or a 3/100 risk or even a 3/1000 risk, and the cost is interference with religious freedom, then the law should be enacted. The parents' freedom to do what they think their religion commands, and what they likely see as being for the religious benefit of the child as well as for their own religious freedom, thus counts for something but not for everything.
There's underniable appeal to this, it seems to me -- but is it just the appeal of compromise for the sake of compromise, or does it actually make moral sense? For instance, how do we figure out whether religious freedom should be valued at 5% of life, 0.5%, or 0.05%? Also, why should the parents' religious views justify the imposition of even modest burdens on the children, who may ultimately have their own views, and who now have no religious views (at least from our secular legal system's perspective)?
In any case, all this leads me to still tentatively think that the practice probably should be banned, if it does indeed pose a risk to infants' lives. But I hope it helps illustrate why the matter is more complex than it may at first appear.
Incidentally, if you're wondering, the reasons that I take a different view with regard to parents' potentially harmful teachings of children -- even when the teachings may eventually (and not just imminently) lead the children into dangerous behavior -- are described in Part III of my Parent-Child Speech and Child Custody Speech Restrictions.