With regard to (1), I don't see a particular distinction between keeping neighbourhoods free of sex offenders and keeping neighbourhoods free of thieves. Thieves are a threat to the property of their neighbours, just a sex offenders can be a threat to their children. Perhaps we might say that children are more valuable than property, but that would suggest that we treat thieves less seriously than sex offenders, but not that we refuse to forbid their presence at all. Perhaps there should be a maximum number of thieves that can live in a neighbourhood.
With regard to (2), I find disturbing the implied doctrine of permanent punishment. If we are to punish some offenders permanently, then it should be those who have committed the worst crimes. Rape, I'll gladly accept, is horrific. Child sexual abuse is also bad. However, murder seems to be worse or at least as bad as child sexual abuse. So, if we are to punish the worst offenses permanently, and we consider child sexual abuse amongst the worst offenses, then we should consider murder the same way. But, we do not. We allow murderers to be paroled, and to be released to full freedom after having served their sentences.
Overall, then, there seems to be a disproportionate treatment of sex offenders in comparison to other crimes. Which suggests policy should develop in one of two ways: either towards stricter treatment of non-sex crimes, or to looser treatment of sex crimes. The former would institute a punishment-based model: in a slogan, if you do the crime, you do the time. That is, every offense deserves punishment, for punishment is the only way to deter and correct criminal behaviour. The latter, by contrast, would institute a rehabilitation-based model: in a slogan, everyone gets a second chance. That is, although offenses may deserve punishment, punishment is not always the best way to deter and correct criminal behaviour.
I would tend to favour the latter approach as a general approach. For one, crime can sometimes be a mistake. We have all heard the story of the child (or teenager) who fell in with the "wrong crowd" and performed actions that they later sincerely regretted. It would be blatantly cruel and unfair to punish this child in order to deter behaviour, given that the presence of sincere regret (and shame) is often enough to accomplish this goal.
For two, even if crime is deliberate, punishment can incubate criminal attitudes. We all know that prison is very often a haven of vicious conflict and criminal behaviour (assault, murder, rape, etc.). If we expose criminals to this environment, it is, it would seem, at least as likely that their current criminal tendencies will be hardened rather than undercut. Perhaps this would suggest that prison should be reformed, but I have a hard time understanding how one could do so without, again, reinforcing criminal attitudes. If the violence between prisoners is controlled, then it seems a hatred of authority (who would have to enact pretty severe controls) would be supported in the prisoners. A hatred of authority, if a reflexive disposition, would extend to police as easily to prison guards. In short, the punishment approach cannot work: this approach assumes that, in effect, unwanted attitudes can be "beaten out" of someone. But human psychology is far more perverse than that -- the response to violence is often increased violence, the response to paranoia is often increased paranoia, the response to distrust is often increased distrust. Indeed, one can see this with animals as well: an animal who is beaten becomes unfriendly and violent, while one whose good behaviour is rewarded (and bad behaviour ignored) becomes sociable and a pleasant companion. Insofar as we are animals, this feature exists in human psychology and must be respected. Hence, the rehabilitation-based model is to be preferred.
Which, after a long-winded exposition, suggests that sex offenders should be treated, monitored as much as is necessary to ensure the treatment has stuck, and allowed as much freedom as any other citizen once it has been demonstrated reasonably that they are unlikely to reoffend. To be sure, this is no guarantee that reoffenses will not occur, and a system which permanently punished sex offenders could (at least in theory) offer this guarantee. However, it is important to note the costs involved. Many offenders who would not reoffend -- who have "learned their lesson", so to speak -- would be subjected to the same harsh treatment as those few genuine hopeless cases. So, we would be treating unalikes alike -- using the same approach on those who will lead harmless lives as those who will not -- which violates one of the most basic rules of justice. Moreover, we would be trading off the suffering of the offenders against the benefit to society at large -- in other words, additively comparing the losses of one group to the gains of another -- without regard for how illegitimate it really is to trade people around as if they are nothing more than empty vessels for loss and gain. That is, we would be violating the essential autonomy of persons in order to produce greater social benefit.
While these may seem like vague abstractions when compared to the concrete suffering of a child, it is only by attention to these vague abstractions that we can keep from being diverted by our disproportionate concern for our immediate emotional reactions to suffering. That is, it is only by attending to the principles underlying our proposed policies that we can ensure that our policies are genuinely fair and just. Otherwise, we run the grave risk of treating those we find personally repulsive in a way that we cannot justify morally.