A few suggestions:

Some topics to have focused discussions on:

0. Advantages/disadvantages of SAA approach when compared with purely empirical algorithms. Why are we going this way?
1. Appropriate steps to obtain Rrs/Lwn to invert from level 1 products.
2. Assumed relationship between Rrs and IOPs. 
3. Fundamentals of inversion procedures: e.g. mimizations vs. matrix inversion.
4. Assumed eigenfunctions of IOPs. The science behind the assumed shapes.
5. How to obtain uncertainties in inverted IOPs.

All of us have opinons pertaining to all these subjects. Depending on our numbers we can discuss them as a single group or breakup to small groups and report back (e.g. an hour of discussion within the small group and then 5min presentation of recomendation to the greater group).

Cheers,
   Emmanuel