So there are 3 primary models that have been used to value a bitcoin:

1. Network-based (like this)
2. Production-based (where you dig into the cost as you mentioned - see Adam Hayes) - the largest issue with this is that huge assumptions need to be made in regards to the cost of electricity, and other things. These are big unknowns and you would be making, at best, close guesses. I have tried and have the work if you're interested. So the results can easily be fitted based on these assumptions
3. Its value as a currency, and supply.

The first and the last have been the best way to go. Energy aside, if people are supporting the network via transactions and security (mining), we can tie a value back to it. In a social network, if nobody is online, the network has no value, as the people are the commodity. Much is true here, where the coin is the commodity, as a general premise. # 3 suffers from the opposite of what #2 does, but still comes closer than #2. We lose sight here from a macro not a micro perspective. Chris Burniske is the source of #3.