Take home idea: we really don't need smart contracts. The Oracles cause centralization issues, and while smart contracts sound sharp, I really don't see a reasonable use case outside of finance/law (where the Oracle would be subject to regulation, and therefore would have the proper motivation to behave) 
On the issue of Oracles, I also had an epiphany on that recently:
Originally I couldn't envision many valid use cases for this new craze building on top of more powerful block chain scripting and the block chain as a generalized state transition database, i,e. the block chain + scripting as a Turing complete machine.
And that was to a large extent because data feeds on external events break consensus algorithms. But then I realized that external events can be voted on.
Any contract that requires an external data feed breaks the ability to have a valid consensus algorithm. So this limits us to scripting which refers to data that is already on the block chain, i.e. the only thing that a block chain can validate are state transitions from an initial set of data, i.e. the genesis block.
So this basically limits what block chains can do, to financial contracts that involve how value is transferred over time and voting. Our contract logic can't refer to events that occur external to the block chain, except by voting. So this means external data feeds (events external to the block chain) can only be accommodated as voters, i.e. if one reporter of the event is authorized then it is a 1-of-1 quorum and if there are 5 reporters of an external event, then say our contract requires 3-of-5 to agree on the report, and then of course the contract has to have logic for what to do in the case that the quorum on an external event can't be achieved.
...
And that was to a large extent because data feeds on external events break consensus algorithms. But then I realized that external events can be voted on.
Any contract that requires an external data feed breaks the ability to have a valid consensus algorithm. So this limits us to scripting which refers to data that is already on the block chain, i.e. the only thing that a block chain can validate are state transitions from an initial set of data, i.e. the genesis block.
So this basically limits what block chains can do, to financial contracts that involve how value is transferred over time and voting. Our contract logic can't refer to events that occur external to the block chain, except by voting. So this means external data feeds (events external to the block chain) can only be accommodated as voters, i.e. if one reporter of the event is authorized then it is a 1-of-1 quorum and if there are 5 reporters of an external event, then say our contract requires 3-of-5 to agree on the report, and then of course the contract has to have logic for what to do in the case that the quorum on an external event can't be achieved.
...