Sounds like a good rant: "Mobile game guru Greg Costikyan, formerly with Nokia Research but now head of Manifesto Games, highlighted a number of problems facing the mobile games industry, and developers in particular."
...but, reading Carlo's notes I'm struck that most of the problems Greg has highlighted aren't problems with the mobile games industry, but problems with the mobile games industry if they want to sell the kinds of things the console games industry sells:
1. Consumers don't necessarily buy games based on a single line of text. Many of the channels for selling mobile games (late night TV ads, press ads in the back of Viz or the back cover of Puzzler magazine) allow fuller descriptions of games. Even sales of games via operator portals usually include screenshots.
2. "Mobile device will always be an inferior game platform". Sorry, I don't buy that; or has my chess-set been obsoleted by my Gamecube because it's only got a resolution of 8 x 8?
3. J2ME can access phone and date book information. Not all handsets can, but hey - you're going to be building different versions of your game for different devices anyway, and at least the mechanisms for address book access are reasonably standardised in JSRs.
4. You can add new levels and content to a Java game on the fly. Flirtomatic is a great example of a massively networked mobile application. Sure, this brings its own set of problems, but it's eminently possible, and has been for some time (we wrote Java games that download new levels back in 2002).
5. Super-distribution of the kind Greg describes can be done. We're doing it on a project right now. It's a little bit fiddly, but it's feasible.
But overall Greg seems to have it nailed...