When you force children to do things you think are good (or to not do things you think are bad), that means that any irrationalities you have will be passed on to them (because they can't properly try out dissenting ideas).
If you're wrong, and you force the child to do whatever it is regardless, it will be hard for either of you to find out, and so it will be hard for the child to make progress beyond your current ideas. Not just that, but it will be hard for either of you to find out whether you were right, too.
For example, in economics, if we make something government-funded before trying it out on the market, we won't know whether it would have been a success or not. Maybe it was a great product and it would have sold a lot. Maybe it was a terrible idea and would have flopped. If we just subsidise it, we can't tell either way.
Ideas have consequences. The idea that you shouldn't force kids to do stuff means you shouldn't force them to: brush their teeth, go to school, not play with fire, be social, not play on the roof, do chores, not watch TV, go outside, go to bed, learn to read, eating 'healthy', sharing toys, exercise, etc., etc. This means that if they want to do these things, they should be allowed to.
What you should do is offer advice. If your reasons for doing or not doing whatever it is are good, the child will want to follow the advice. If the child still doesn't, that means he has unanswered criticism. Forcing children to do stuff they don't want to do is an anti-criticism approach (which is very bad, because criticism of flawed ideas is the only way to make progress to less flawed ideas).
Related link:
* Don't expect people to disagree
I think most people do under-use advice and over-use coercion, but equally I think your reasoning is unsound: this is the crucial premise;
"If your reasons for doing or not doing whatever it is are good, the child will want to follow the advice."
If a child uses hyperbolic discounting, then something could be in their best interests but they still choose not to do it. Our brains were evolved at a time when hyperbolic discounting was basically always the best heuristic, but that's not the case now.
Obviously, this applies to adults too, but most adults have already learnt the uncontroversial ones, and things like regulatory capture, Hayekian information issues, etc. prevent paternalism from being a good idea.
http://en.wikipedia.org/wiki/Hyperbolic_discounting
Wanting to do something that has benefits now but bad/worse stuff later is just a mistake. This mistake can be explained to the person making it, and if it's done persuasively enough, they won't want the now-thing anymore. Rather, they'll prefer to wait.
So, when giving the advice about what to do, you might also include information about this mistake. The child's criticism is something like 'why should I do what you say when I can get this benefit now if I do this other thing?'. If you don't address that criticism, that's not a full argument (or at least, it's not a persuasive one to that person).
I think you under-estimate how hard it is to make people act on this knowledge; hyperbolic discounting is rife, even among adults. I understand it and am a fairly driven person, but still do it, so I'm dubious as to how effective persuasion can be on the average child.
Do you have any studies to support your POV?
Why do you want to make people act on this knowledge? I mean, I understand helping someone internalise it more when they want to make choices better in this regard. But to make them act on it sounds like force instead of persuasion.
If you struggle with acting on this knowledge, that's a sign you haven't fully internalised it, or haven't fully been persuaded by it (which is basically the same thing). In other words, there's something you don't know about it which, if you did, you wouldn't make that mistake. There's a misconception or problem somewhere in your mind that, if addressed, would mean you would have no trouble acting on it.
Why would persuasion be harder on children? It's usually easier, because they tend to not know as much.
What sort of study do you mean?
If we're saying someone's only been persuaded of something if they can overcome hyperbolic discounting, persuading anyone of everything would become very hard. I think forcing them to behave sensibly, and hoping they acquire good habits, is a lot easier.
The issue, of course, is whether or not the improved efficiency for passing on useful ideas makes up for the cost of not improving them so much. Do you know of any research that suggests one way or another?