Recently, the new gTLD go-nogo analysis has focused on the single question: "do the economic benfits outweigh the costs?" On the surface, this question makes sense. If it were possible, using a field of study such as "economic analysis" to determine the answer to this question, then the deciscion to roll out *unlimited* new gTLDs would be simpler. ICANN has spent close to a million dollars commissioning "economists" to answer this question. Seven "economic studies" have been comissioned, and in seven out of seven cases, the result of the study has been inconclusive. The costs notwithistanding, these so called "experts" have only reached one conclusion: "it appears that the benefits outweigh the costs, but we can't be sure". As a Stanford PhD myself, and as a participant in the last two years of the ICANN gTLD formulation, I think I can improve on the findings of these highly paid experts. But this first requires us to consider a radical belief -- that the future is not determined and knowable, but rather is probabilistic. It is certainly possible that the new gTLD program will be be a complete bust and that very few registrations will take place. It is also concievable that new gTLDs will be highly sucessfull, and that in aggregate the new gTLD rollout will offer a competitive choice to .com and major cctlds. To understand the risks and benefits of the new gTLD program we need to analyze each of these two scenarios. In the "bust scenario" the new gTLDs will not, on aggregate have a large consumer demand. To quantify things. lets put an aggregate number of 5 million on the total number of registrations accross all new gTLDs in this scenario. Is this scenario plausible? Several of the so called "studies" seem to think so, based on the extremely low numbers of .museum and .areo, as well as TLDs such as .mobi and .biz which, although succesfull, have not lived up to their initial projections. However, plausible or not, the economics of this scenario are as follows: under 5 million total registations, implying probably less than 500 thousand (as an extreme upper bound) of defensive registrations by brands. Using data from the Minds and Machines UDRP analysis, this would correspond to about 500 UDRPs. Between these two costs, the costs to brand owners under this scenario is under 10 million. What of the benefits? In this scenario, they are few and far between. Possibly certain groups and geographical entities will get their own gTLDs, but given the low uptake in the scenario, it would be hard to concieve of a large amounts of benefits. So, to recap, in the scenario in which the gTLD program is unsuccesfull, the costs are (relatively) low, and the benefits are also low. The costs might very well outweigh the benefits, but both numbers are small enough that the only conclusion would be that ICANN had wasted its time with gTLDs. So now let's consider the "success" scenario. To pick a number lets define this as 50 million or more registrations for the entire new gTLD program. Is this number crazy? There are currently 200 million domain names at the second level. A 25% increase allowing for all new cities, states, communities, and generic terms in all languanges seems certainly concievable. In this scenario, its hard to argue that costs exceed benefits. If 50 million people choose a new domain name product, then, by definition, there is consumer demand for that product. Certainly the costs to brand holders will increase, but again, using the Minds and Machines UDRP study, the best case for this cost increase would be 25% -- ignoring any effect of the new URS regime. So, to recap, either the new gTLD program will be a failure in which case ICANN and trademark holders really have nothing to worry about -- or it will be a success, in which case the benefits will clearly outweigh the costs.