Thanks for all the helpful guidance. I'll work on getting the JSON
implementation updated before releasing it.

I don't know if it's of interest but the Twitter firehose (as deliverd
by Gnip) is line oriented JSON. Each line is a tweet and all its
metadata. This format is handy for doing things like counting the
number of records with a 'wc -l' instead of having to parse the
JSON...which can be expensive when there can be 10M an hour.

//Ed