Resolution:
Practically you can go through the following decision tree:

If you want the query to return the matching tuples as they were when the query was issued, even if the tuples get deleted or modified after the browse call return but while you are iterating through the result set, then you want temporal isolation and then  need to use SNAPSHOT.
If the data does not change or if you are OK getting the updated values then you can use either CURRENT or SNAPSHOT. In that case it is better to use CURRENT if the selectivity of your query is low.

If you want to iterate over the whole space, or over most of the records in the space, even if the result set if your query is large then using CURRENT is better because it will require less memory to be used by the seeders and in some cases it will actually be faster.

You have to write your code such that if you get an exception during the browsing,because of re-distribution,  then you should retry the query.

The difference between BrowserDef.TimeScope.SNAPSHOT and BrowserDef.TimeScope.CURRENT

Refer to AS 2.1.2 HF-10 read me file for more details.