About Spark UnsafeShuffleWriter

I have two questions about the UnsafeShuffleWriter UnsafeShuffleWriter

will be used when all three of the following conditions are met:

• A shuffle dependency does not indicate an aggregation or output ordering.
• The shuffle serializer supports wrapping of serialized values ​​(currently supported by the special serializers KryoSerializer and Spark SQL).
• Shuffle produces less than 16,777,216 output sections.

I am confused about the first two conditions.

• Why would a random permutation not indicate an ordering or an output ordering? I find it a good idea to use UnsafeShuffleWriter
  
  if mapSideCombine = false, regardless of whether aggregation or ordering is specified.
• Why does the serializer have to support wrapping the serialized values ​​where movement will be used?