There's a common scheme used in discussing children's rights, going back to Joel Feinberg. There are A-C rights, which are shared by adults and children. There are A rights, which only belong to adults (right to vote is the usual example). Then there are the somewhat misleadingly named C rights, which are rights that are had by children as the norm and by adults only under special circumstances. Feinberg notes two subclasses of such rights, in particular: dependency rights, which are rights to the instrumental goods of life connected with the fact that the person in question is a dependent, and what he calls "rights-in-trust". I would suggest that "rights-in-trust" is an ill-formed category, and that its influence on discussions of children's rights has been detrimental to the discussion as a whole. I will mostly focus on laying out some basic reason for skepticism about rights-in-trust, and take it as given that if there's reason to think the category ill-formed that the confusion it can cause to arguments down the line is itself reason to stop using it.
It's difficult to get any adequate characterization of what these rights-in-trust are supposed to be. Feinberg says that they are analogous to the kinds of A rights that are associated with autonomy. (I don't think it's usually realized just how high this sets the bar for formulating it; autonomy rights have to be very precisely formulated, because small divergences in formulations, or interpretations of formulations, can lead to extremely different practical conclusions.) The basic idea is usually said to be (and is explicitly said by Feinberg to be) that children have autonomy rights but cannot exercise them, so we 'save' these rights for them so that they can exercise them later. Both clauses should worry us a bit.
In the first clause, which gives rights-in-trust their occasional name of "anticipatory autonomy rights," we are quite literally talking about the autonomy rights of the non-autonomous, rights that people somehow have now that we are somehow morally justified in not letting them exercise now. It's fine if you don't think this is a contradiction; but it needs to be more widely recognized that it is a paradox in need of a precise account to show that it is not a contradiction.
In the second clause, this talk about 'saving' rights is itself troublesome, because it's unclear what counts as 'saving' a right for someone for later. Rights are not like money or heirlooms, which are very limited kinds of trusts. There is very little you can do with them, so what counts as good stewardship of them for later can be quite precisely defined. Every genuine right, however, covers an indefinite scope of action within a more-or-less precisely defined domain of action; what you can do with a right is limited only by human ingenuity in using it. So the usual way to determine what counts as good stewardship or conservation is useless here: everything you do with everybody can possibly affect the precise future actions that are available under a given right. If we posit that I have a right to choose my own religion, just as a toy example, the fact that American schools do not teach about Sikhism in a culture that is largely oblivious to the fifth largest religion in the world is limiting the ways that I would be able to exercise my right. This is obviously not going to be avoidable. And the same problem arises with other rights that are not such toy examples. The maintenance of political parties massively limits the scope of how I can use my right to vote, for instance. Thus it would seem that it would have to be the right itself, not any particular way of exercising it, that is to be kept in trust. But how one protects a right whose exercise is some set of unknown actions in the future is somewhat problematic. If I have a right, right now, we can just look at what the scope for using it is right now, and protect the right by protecting that scope of action. If there is question of whether a greater scope of action is allowed, we work it out by discussion and like, using the current scope of action as our basis. Even the right to vote, which superficially looks like it can only be exercised occasionally, in reality also guarantees that precise mechanisms and institutions be in place so that we can vote at a designated time, and this aspect of the right can be exercised at any time. You have the right now, you can exercise it now by demanding protections under it now, and what protections you can demand are very precisely definable. This is simply not the case with autonomy rights; ex hypothesi, they are not rights that children have except 'in trust', and there are no precise mechanisms and institutions within which they exist because the whole point of an autonomy right is that it follows from human reason itself, which covers the whole of human life. Beyond protecting life itself, how do you judge beforehand what needs to be done to keep autonomy rights exercised?
This is related, incidentally, to an immediate practical problem with the idea, namely, that it's unclear that any parent can actually parent in this way. Parents are usually trying to deal with immediate problems in the present and, to the extent that they are able, set their children up for success as best as they know how. But protecting a currently existing autonomy right is very difficult -- it requires having the right kind of government, the right kind of culture, and so forth. These are things that parents can't guarantee, so are they not the trustees of these deferred rights? But if society as a whole holds these rights in trust, then it's unclear why this doesn't require an extraordinarily high level of intervention in parenting by the various means by which society protects rights. Our trust-the-parents-unless-there's-a-sign-of-a-serious-problem approach would be as absurd as just trusting that everybody has autonomy rights until it's obvious that they no longer do. Likewise, looking at the discussions, it seems like there's a common assumption that this standard would favor progressive rather than conservative parenting, but without a precise account of what it is to hold these rights in trust, it's not obviously congenial to either; indeed, without such an account we have no idea at all what parenting would have to be like to take this as a major standard of parenting. Just protecting the (much more precisely defined) dependency rights of children is full-time work for parents; even trying to set their children up for success is an icing-on-the-cake thing that parents are usually relieved to be able to manage to the limited extent they can. How they are supposed to take their children's entire future into account, in some vague open-ended way that can take into account all the situations that are actually possible, and do it all the time, is a problem I don't think can be assumed solvable for most parents, regardless of their parenting style. To put it in other words: the difficulties of protecting autonomy rights seem to transfer to these rights-in-trust -- but none of the ways that make it possible to get around these difficulties when dealing with autonomy rights actually had right now (e.g., the fact that you don't have to cover indefinitely many situations but only the one in front of you right now) seem to transfer with the difficulties. You'd have to have other ways of handling the difficulties -- and without a precise account, it's unclear what they would even be.
One of the common ways of talking about these rights-in-trust is to say that children have a 'right to an open future'. Feinberg himself, despite using it as the title for his famous essay on the subject, regards this as a loose, secondary label for these kinds of rights; it gives you the gist, and so is accurate enough, but there is no unrestricted right to an open future. It's just a way of distinguishing between autonomy in a proper sense and whatever it is that children have that corresponds to it. If you look at later discussions, though, it's quite clear that it's the set of ideas associated with the label, right to an open future, that has really dominated the discussion. This is worrisome in itself. I don't think Feinberg ever does manage to give us a sufficiently clear notion of rights-in-trust, but at least the point was to be relatively precise; 'right to an open future' was just a handy, imprecise label to make it easier to talk about the rights in question without confusing them with autonomy rights in the strict and proper sense. Obviously there's no way to determine what an 'open future' will be beforehand; and keeping some futures 'open' always involves trying to close down others. (Claudia Mills has a good discussion of some aspects of this in her article, "The Child's Right to an Open Future?") Again, we need the precise account, not the label, and especially not the set of mere associations that the label misleadingly is capable of suggesting.
My suggestion is that, in fact, A-C rights (human and civil rights shared by adults and children) and dependency rights are quite enough. That is, it's not clear that there is anything gained by dragging in these 'rights-in-trust'. What's more, I think trying to fit rights into such a category actually deteriorates serious rights that are held under these two other categories. One of Feinberg's examples, for instance, is the right to exercise one's religious beliefs. According to Feinberg, this is an A right that is only held in trust for children, and one of the whole points of classifying them this way is to insist that the latter is a very different thing from the former. But in fact, the right to exercise one's religious beliefs is clearly an A-C right; the difference is not in the right itself, and there are lots of situations in which we would obviously recognize that the child already has the right to exercise his or her own religious beliefs. If someone were to try to force your child to join (to use a purely random example that should not be taken as in any way a likely possibility) Tenrikyo, and you are not yourself a participant in that religion, this isn't just a violation of your rights as a parent, but a violation of your child's own religious rights right then and there -- not of your child's rights down the road as an adult. What is really happening here is that the same right has different modalities because the right doesn't eliminate the facts of the parent-child relationship, which includes education in religious matters. Everyone outside that relationship, though, still obviously has to respect your child's basic right to exercise whatever religious beliefs he or she might have. The child's autonomy right may be more limited than the adult's, because the child's autonomy is itself only being developed, but it is not nonexistent! The category of rights-in-trust downgrades the real, however sharply limited they may be, autonomy rights of children by treating children as if they had none.
Feinberg himself comes close to recognizing this last point. He recognizes that it's hard to draw the line when dealing with older children -- he just doesn't properly consider that all the lines we draw between childhood and adulthood are arbitrary lines of convenience, so the most natural conclusion (which is the implicit assumption in how parents in general actually parent, I must point out!) is that the rights grow naturally with the natural growth of the child. If that's the case then, many of the things that would be categorized as 'rights in trust' will in fact be genuine A-C rights at different stages of growth.
It's also unclear why anything else couldn't be taken care of as just ordinary dependency rights. All the most obvious ways in which parents and, for that matter, society, can take the future of their children into account are things like feeding and educating for which the child has a dependent's right here and now. All the alleged conflicts between rights-in-trust and the A rights of parents that end up having to come before the court are much more naturally seen as apparent conflicts between the A rights of parents and the dependency rights of children. Feinberg tries to handle this by saying that these conflicts are often not dealing with the immediate health or development of the child; but I don't see that this is anything other than an obvious modal scope fallacy -- having the dependency right in the present doesn't mean that it doesn't cover more than the present.
So those are some of the reasons I'm skeptical of this whole notion of rights-in-trust. There are attempts to answer bits and pieces of these problems, but I find none of them adequate. It just doesn't seem properly thought out. And given that it has dominated so much of children's rights discussions, I think this is a very definite problem.