Hey Curt,
Most of my own runs, using the 2 small VM default, resulted in 3 normalized hours of usage, which equated to around 25 cents per run.
how much data was that?
The cool thing about running this job inside Elastic MapReduce right now is the ability to get at S3 data for free, and for cost of access outside of it, both pretty reasonable sums. Right now, you can analyze the entire dataset for around $150, and if you build a good enough algorithm you'll be able to get a lot of good information back.
We're working to index this information so you can process it even more inexpensively, so stay tuned for more updates!