Thanks for the reply - I think that that is the point of utility in my opinion. Like you said with your air example, utility alone is worthless. I am going to be working with some ICOs on their token economics, and the key thing that people don't seem to realise is that the only goal of an ICO regarding its price target is to create a bottle neck system. This is composed of a utility, that is complimented by a sensible yet lucrative supply and demand structure.
To take Bitconnect, for example, it did have value. All ponzi schemes and pyramid schemes have value while the scheme is sustaining itself, as their value is the monetary velocity they create, which coupled with volume creates what I like to call monetary momentum (this will be one of the next big terms for tokens in my opinion, I think I'm the first to have used it). They have a value in the sense that demand keeps on going up, as does its utility (momentum) - this creates a bottleneck, but the issue with schemes like Bitconnect is that when the scheme crashes, so does all of its momentum. If momentum is at 0, when you times momentum by utility value (although assigning a figure to utility is difficult), you essentially get zero.
I think things being overvalued and undervalued is a simple matter of smart economics - it's something that I have had to learn myself, and have not been able to find a book that teaches me it thus far. I have been considering writing a 20-30 page article all about evaluating tokens, how it can be done etc. and using past tokens to showcase to some extent that my methods work. My main issue at the moment is finding the monetary support for such an article - it would take a couple of weeks to write at least, so getting financial support to not work on other projects is difficult, but a way around it that I have been thinking of is to post it in snippets on Steemit - I would really prefer to have it as an e-book and then break it up on Steemit though. Hopefully it could be used as a guideline for ICOs. Let me know what you think, I'm pretty analytically minded so I like to try and apply models to everything, which doesn't always work - I think it does in this case though.