Many of the analogies used to explain Bitcoin are, unfortunately, misleading. Saying 'each bitcoin', for example, as if it were a tangible collectable, introduces a bunch of associations that aren't helpful for understanding.
Instead, think of Bitcoin as a giant consensus accounting ledger. Everyone can see which crypto-signing-keys have the right to spend which balances. (The balances happen to be measured in an integral unit called a 'satoshi', and 100,000,000 of these satoshis equal "1 BTC"... but these are tallies in a ledger mapping key-identities to available-spending-balances, not rare collectables.)
Given that, a person's Bitcoin holdings are simply numerical totals which, at the direction of their signing keys, can be handed in full or part to other signing keys. But it works like having a bank balance at a shared "bank in the cloud", massively replicated, with no single authority or point-of-failure, where everyone can see every transfer. (Balance transfers are public announcements, not the secure sending of a rare secret.)
So bitcoin balances are valued with regard to $USD (or other currencies) like other foreign-currency bank balances - who's willing to buy it, for how much, compared to who's willing to sell, for how much. That is, yes, "supply and demand", but based directly on the willingness of holders/seekers to transact, and only indirectly on the total 'supply' in existence.
That the price has run up indicates that holders of current bitcoin balances want more $USD to give it up, and that seekers are offering more $USD to get it. Every day more people hear about Bitcoin, and slowly and slightly (because it's pretty complicated and different) come to understand it better. Also, using a heuristic common for new technologies, people only believe something new will survive after seeing it survive for a while. [1]
From my perspective, heists and scams are good for Bitcoin, because they're providing a trial-by-fire and accelerated-education in various economic, security, and usability topics for anyone involved (or just watching from the sidelines). Things that took decades or centuries to work out in traditional economic systems will be resolved in months or years, and with countermeasures optimized for today, not the 18th/19th/20th centuries.
[1] See for example Nassim Nicholas Taleb on the 'Lindy Effect': http://www.wired.com/opinion/2012/12/worlds-not-ending-but-t...