there were *a lot* of duplicate events at various times. the firehose was re-broadcasting itself at multiple times the normal rate for a while. oops!

we also re-ran events for a couple hours from a couple PDS instances to ensure all events came through on the main firehose