I can't help but also think of a 2+2 post I read recently when looking at this discussion which was exactly on the topic of value betting.

The important thing is to understand when you are betting for value. Simply betting out whenever the action is on you is not a value bet. Simply betting out when the action is on you and you have a piece of the board is not a value bet. Simply betting out when the action is on you and you have TP or better is not a value bet.

A bet is a value bet when you have at least a basic understanding what kind of hands your opponent can have and expect that you will be called by enough hands that you beat that betting out in the first place is +EV.

Bet sizing is a wonderful discussion to tack onto a discussion of value bets (thanks Spenda). A basic princple here is that the villain might not call with worse very often if you bet big - but might call with worse if you bet a bit smaller. This results in using a specific bet size to manipulate the villain's hand ranges to contain hands that are weaker than yours and from which you get additional value.

When thinking about bet sizes there are some important considerations.
What is the biggest bet size that much weaker hands will call?
What is the smallest bet size we can get away with betting for value without seeming weak and being bet into as a bluff?
What is the biggest bet size that our opponents second-best hands are likely to call?
What bet size is likely to fold out hands stronger than yours?
What bet size will fold out too many hands weaker than yours that you are no longer betting for value?

Often you can define a bet size range that will have approximately the same behaviour for a villain and you can decide what amount to bet inside that range to optimise your EV.

And as Spenda pointed out position can also need to influence your bet sizes. When you are OOP you often need to bet something like TPTK three streets in order not to give up the initiative and be pushed off the hand where in position you would prefer only to bet two streets - with the implied logic that if you end up playing for three streets you are typically not a favourite to have the best hand. By betting smaller OOP you may achieve a level of pot control so the final pot is closer to where you want it to be while also manipulating the IP opponents hand range to contain more hands that you beat making it overall more profitable.

It is absolutely crucial to understand why you are betting and what you are trying to accomplish with your betting. Not understanding why you're betting is like being approached by a stranger in the street who tells you: "If you'd like to gamble give me $10" - you give him $10 and he tells you: "Shame, you lost." The fact that in poker you will often win when you bet for no reason and have positive reinforcment as a result does nothing to detract from the validity of the outrageous example. You don't know if the bet is +EV or -EV and yet you bet.