With the release of 2.0, we have the possibility of significantly improving BSON encode/decode performance in many cases. Some internal teams use this ReflectionFreeDCodec that reduces BSON encode/decode time by about 50% for supported types. With the simplifications to the codec system in 2.0, we can borrow some of the logic from the ReflectionFreeDCodec to get similar performance in the default BSON library. That's a performance benefit for everyone, and means nobody needs to maintain the ReflectionFreeDCodec anymore.