I've addressed the same question at some length in Gambling for the Good, Trading for the Future: The Legality of Markets in Science Claims, 5 Chapman L. Rev. 159, 172-76 (2002), and more briefly in Prediction Markets for Promoting the Progress of Science and the Useful Arts, 14 Geo. Mason L. Rev. __ (2006) (forthcoming). Jason refers only to the later paper, which relies on the earlier one for detailed analysis. That perhaps explains his complaint that, "Bell  does not cite a relevant part of the [Commodity Exchange Act's] definition of 'excluded commodity' . . . ." I do cover that topic; I simply don't do so in the paper that Jason relies on.
What do I say in that earlier paper? First off, consistent with my later paper, I argue that the CEA does not give the CFTC jurisdiction over run-of-the-mill prediction markets. "The CEA does not cover contracts intended to effectuate future delivery, much less contracts that effectuate immediate delivery." 5 Chapman L. Rev. at 170-71 (footnote omitted). With that in mind, it becomes obvious why prediction markets should generally fall outside the CFTC's jurisdiction. Such markets typically deal in the present delivery of conditional rights: claim coupons payable in the event some contested claim comes true.
As all my writings on the topic have taken pains to concede, however, defining the exact scope of the CFTC's jurisdiction proves frustratingly difficult. A court might well decide that my argument smacks of logic-chopping. The Gambling for the Good paper thus spends a fair amount of time exploring what would transpire if the CFTC did win jurisdiction over prediction markets. Specifically, pages 172-76 discusses whether the claims traded on a prediction market would qualify as "excluded commodities" and, if so, what sort of regulations would apply. Here's my bottom line:
[S]hould a market in science claims find itself subject to the CEA, it could attempt to qualify for the “excluded commodity” loophole that would largely free it from CFTC regulation. It looks highly probable, however, that the CFTC would have wide discretion to thwart any such attempt. At the least, to judge from precedent, the CFTC would probably not exclude a market in science claims from its regulations without also imposing crippling conditions.
Friends of prediction markets should not despair, however. First, as I've said, there remain good reasons to deny the CFTC jurisdiction over prediction markets. Second, the "hybrid instrument predominantly a security" category, described in Prediction Markets for Promoting the Progress of Science, offers yet another promising but untried legal hack for both preempting state gaming laws and forestalling CFTC regulation. Third, various state law pertaining to games of skill may well provide shelter for prediction markets.
That last legal strategy remains, so far as I can tell, completely unexplored in the extant literature on prediction markets. My initial research on that topic has proved heartening, encouraging me to learn more. I'll plan to say more about the topic later.
[Crossposted to Midas Oracle.]