I'm certain you're right but we need to test that :)
the way I read OP was the abandon rate was ~100% previously in areas with high latency/low bandwidth, and his experiment had a non-zero abandon rate that was shrinking as word spread. Perhaps I need to read it more carefully.
the way I read OP was the abandon rate was ~100% previously in areas with high latency/low bandwidth, and his experiment had a non-zero abandon rate that was shrinking as word spread. Perhaps I need to read it more carefully.