The debate is mainly about the future and how the increase in transactions will be handled by the network.
As blocks fill up, solutions need to be made.
Gavin wants to increase the maximum size of the blocks, to allow more transactions. Opponents don't want this, because they fear massive storage size needs and slower mining in places with high latency (china).
One important side node: The devs from the dev team that strongly oppose are heavily involved in creating a sidechain network company to fix this. It seems that financial motives play a big role too.
It's not as partisan as you make it out to be. While I do not support BIP 101, I do support BIP 100 and BIP 102. This is not simply about those who want bigger blocks and those who oppose. That's part of the problem of lumping us into partisan camps (like XT vs. 1MBers).
Some of us think BIP 101 is rash and lacking in foresight regarding future problems. Personally, I think it's incredibly audacious to assume that we can -- today -- plan for what real adoption will look like 20 years out and arbitrarily putting it at 8000x current capacity, with sufficient thought for hardware and bandwidth capabilities, security in the case of such a potentially bloated chain, and many heretofore unknown and unacknowledged problems that may arise. And that's whether or not Gavin, Hearn and the developers of today that audit the code have the technical foresight to understand and plan for every contingency. To be clear, that is impossible. (Nevermind that Moore's Law is an unproven theory that shouldn't be used as the basis for block size. If the predominant issue appears to be capacity -- and the extent that this matters is dependent on adoption -- why an unscientific theory about expected growth in processor capacity is applicable is beyond me.)
I think people need to lose this idea that this is the "fork to end all forks." That's not how bitcoin works.
I support BIP 100 because it takes a more conservative approach that gives us ample time to observe scalability over time, and to stay in line with real growth in transaction volume rather than some arbitrary vision of exponential adoption seen in BIP 101. I think having the miners vote among themselves as to when a limit increase should occur is a preferable approach. And 90% is a much better measure of consensus than 75% -- which to me is more like 50-60% + a lucky streak.
Consensus is paramount. I believe all of the above are good reasons why BIP 100 is far preferable to BIP 101, and why it will be easier to maintain consensus post-fork.