We define a microprediction oracle as an apparatus that you can reward for a repeated prediction task, on an ongoing basis, whose accuracy is eventually hard to beat for the same or lessor cost. In this module we discuss the implications of this aspirational property.
Using a concrete example of predicting the arrival of a school bus, we reason as to what an oracle might be, in order that it so much as come within a factor of two of meeting the definition. Thereby, we motivate the qualities of this project that set it apart from prediction markets, financial markets, data science contests and asymmetric crowd-sourcing setups.
We hope this helps explain the benefits of an open approach.
To date, no mechanism for aggregating information has met, or even aspired to meet, the key principles of the prediction network that we derive from a straightforward argument.
The distinction between platform and participant should be made as small as possible.
It should be possible for anyone to wield the power of whatever information aggregation ability the platform purports to support. Otherwise, the recursive argument for eventual efficiency fails.