(Sorry for getting to the party so late, it's taken quite a while for me to really formulate a healthy respect for both the good and bad parts of this system).
First off, I agree that TU16 didn't fix the underlying "issue". In my opinion, the losing alliance getting the winning alliance's score is inherently flawed. However, I think this system deserves more credit than we give it.
I won't even bother discussing the issues I find with the pre-TU16 implementation of the ranking system, as I think most of us can agree that is is deeply flawed.
My most direct exposure to the TU16 defined seeding rules was at Cass Tech. Before we continue, we need to agree on the purpose of seeding matches. In my opinion, seeding matches exist to sift through the pack, putting the BEST machines on top (and most especially the best 15, as that's the maximum # of teams who might ever get the chance to decline a selection and still play). If you disagree on this point, then I believe you'll find the rest of my argument invalid (as this is the basis of everything I'm going to say now).
So, do you think that this algorithm brings out the top of the pack? I do. Lets look at the alliance selection results from Cass Tech (I can't say if similar results were seen at other regionals). Straight off usfirst.org...
Code:
1) 217
2) 1718
3) 469
4) 226
5) 2612
6) 1941
7) 308
8) 313
Those were seeds 1 through 8. If you see a large amount of "inter-picking", I think that indicates that this algorithm worked to bring out the top of the pack.
1 Picked 3. Then 2 picked 4. Then 5 picked 7. That is to say, the first 3 picks were all top 8 inter-picks. I think this speaks volumes to this algorithm's ability to sort out the top tier of teams. Granted, Cass Tech was a FiM event with 12 matches, and more matches generally ensures better sorting out of teams. Nevertheless, I believe that this algorithm is effective when sorting out the best teams, especially the very best. If a very good team ends up with a poor win/loss record, but played well, they'll still seed rather high. Likewise, if a middle/bottom tier team gets a lucky schedule and goes 11-1 or 10-2, they're likely to seed lower than a good team that only managed to go 8-4 due to a killer schedule.
So, while I think that getting the winner's points for the losing alliance is conceptually flawed, it helps bring out the best teams to the top of the pack, regardless of the "toughness" of their schedule.
I do think, however, that it does not do a good job of sorting out the middle/bottom of the pack. Nor does TU16 fix the "6v0" condition. However, I think it's a step in the right direction.
Just my $.02.