Conclusion

Because you now know that MIDP 2 is not really an upgrade, you have to ask the question of when it's right to develop for it. Right now devices supporting MIDP 2 tend to be toward the high end of the market, thus they represent a very small portion of the installed player base. Although this will change over time, at the moment the bulk of the market is still very much in MIDP 1 land. For this reason, I make MIDP 1 (usually with manufacturer extensions) my primary target for games and then port to MIDP 2 once completed (if it's worthwhile).

The main reason you'll want to develop for MIDP 2 is to take advantage of the new features that are not available (even with manufacturer extensions). Some of these features are downright cool, such as having your game wake up when network traffic arrives for it. Another reason is the lack of support for many of MIDP 2's features with manufacturer APIs, such as PNG transparency and sound. If your game relies on these features, then go for it.

Finally, due to MIDP 2 implementations covering things like layers and spriteswhich manufacturers can choose to implement using native callsyou can expect games to run faster than MIDP 1 equivalents. Note that I said expect; you can just as easily encounter poor implementations (especially with tiled layers and image processing).

There's no doubt that MIDP 2 is significantly better than MIDP 1; keep your fingers crossed that it becomes the dominant platform sooner rather than later.