We first noticed that one of our Plus users was responsible for a growing number of bad request errors (400 errors) that resulted in degraded performance for the Plus REST API starting at 14:03 UTC, Tuesday, 19 November. We thought we blocked all of these bad requests on Tuesday in order to improve performance. Unfortunately, we learned earlier today that not all of those requests were being blocked - the block we implemented was simply incomplete. As of 16:45 UTC today, we have blocked the remainder of bad requests. And, we continue to monitor.