Tuesday 2016-10-18

Say that Prediction Markets are part one of a two-part process, where the state of the market before the Event happens is what we sample before shipping that sample data to Machine 2.

We want to know the odds that the market is being manipulated. Assuming that's low, we get the wealth-risk-appetite-weighted common expectations for the Event.

So, Machine 2 best(?) uses Events where we don't already have good measures for expectations, and where we have some element of control. We can call this the "Shitty Driver" class of uses. Gov't reps should be chary of "Shitty Driver" since they fear being held to a higher standard / needing to improve their driving. So we avoid this class.

Another class of use is where we have zero control, however we're affected/interested in some way. Call this the "Victim" class. This class is easier to sell as a version of the DEW Line. And there are many(?) under-funded programs in the victim class that could use the help (not sure whether welcome or not).

What's the most under-funded monitoring program?

#predictionmarkets