If you are looking for "update" capabilities when importing, our DataPorter
series product may be of some interest to you (see link in sig line below -
a more advanced update/append function will be available in the upcoming 2.1
series of the product).

However, a few reasons why you're suggestion(s) won't work are as follows:

#1 - " How about an auto setting that picks the most recently modified
version of the contact?

The modified date/time stamp applies to the contact record as a whole and
not to a specific field. However, even if it did, you wouldn't generally
have an applicable field in a CSV file to compare anything to but getting
back to point on the initial reason, you may be importing 10 fields (for
sake of argument) and your contact record may be been changed for something
totally unrelated. In other words, while your contact record may have a
"LastModifiedTime" stamp greater then a corresponding field contained in
your CSV file - doesn't mean that the input data isn't the most current for
for the fields in the dataset.

#2 - "Outlook should recognize that the Contact name is the same & ask which
data for each field to use."

- The assumption in this case is that the criteria that constitutes a dupe
is based on contact name which is not a good field to use exclusive of
anything else (at a "bare" minimum, you'd need something like email address
and even then - this doesn't deal with duplicate contacts that potentially
may exist within a folder)
- If you import thousands of records as many of our customers do on a
regular on-going basis - having Outlook ask for "each" match just wouldn't
be practical.

Once you go outside of "the absolute simplistic" import scenarios there are
a LOT of things that you need to handle - for instance:
    - if you're updating, does an "empty" field mean that there is "no
data" and the original data should be retained OR
      does it mean that the data in the contact record should be cleared
    - what constitutes a field that has "No data" versus one that should be
set to blank (or zero) etc?
       i.e. is a field that has one only one space character (or
alternatively padded with all spaces)  the same as one that is
            "empty"? Are either of these equivalent to a field containing a
NULL value?
    (answer to the above is that there is no "single" answer since everyone
has their own set of definitions and rules
      based on the data that they have to work with and different programs
can generate the same data in different
      ways).

Karl