ICANN is barking up a number of wrong trees with the latest version of its Evaluation and Questions Criteria for generic top-level domains (gTLDs). The document asks for financial projections, the applicant’s goals for its TLD, and the benefits the applicant expects to derive. All that information is meaningless. Instead, ICANN should require applicants to present a detailed risk-management plan, consider price premiums on existing TLDs, and an estimate of value derived from prediction markets.

On page 40, ICANN expanded the financial statements section to require financial projections and their underlying assumptions. At the bottom of page 42, the document states that applicants “must demonstrate a conservative estimate of costs based on actual examples of previous or existing registry operations with similar approach and projections for growth ...” How can applicants quantify conservative estimates? How can they provide any meaningful estimates of growth when they cannot even determine the initial demand for the new gTLD? What we have here is a case of garbage in and garbage out. As a result, viable TLDs may wind up being rejected and nonviable ones accepted.

Some events can’t be predicted, only experienced. For example, what would people pay for apps? How successful would the launch of Twitter be? Moreover, using existing registry-cost estimates may overstate a new gTLD applicant’s cost because existing registrars will have additional economies of scale and scope advantages in operations, marketing, and execution.

Growth and value estimates should be based on prediction markets and price premium estimates among existing TLDs. Although prediction markets are not perfect, they can eliminate a lot of the failures.

You may ask why venture capitalists don’t use prediction markets. The reason is that crowdsourcing would let the competitors of a technology start-up get a hold of strategic information. The situation is different with new gTLDs. Even if an applicant reveals an interesting new gTLD, it need not reveal the outcome of crowdsourcing.

TLD price premiums provide valuable information about which TLDs have fared better than others. Such information can be valuable to applicants who are able to infer the causes of success.

On page 11, the guidelines now require applicants to justify their proposed TLD’s goals and benefits. However, competitive strategy requires applicants to differentiate their products to have a chance of success. They don’t need ICANN to tell them how to make sound business decisions. With the possibility of a thousand new gTLDs, market solutions become more viable than ICANN’s beauty contests.

To decrease applicants’ failure risk, ICANN should require each new gTLD applicant to provide a risk management plan. The plan, which would be in addition to the already approved financial risk-management requirements, would provide details on how the applicant intended to respond to technical failures and registrations that were considerably higher or lower than expected. (The risk range parameters can be quantified.)

In short, ICANN should focus less on applicants’ tactics and more on their risk management strategies. ICANN should make it mandatory for applicants to use prediction markets and should not require applicants to state the benefits of their gTLDs. Only markets can determine the value of a gTLD and the value of the new gTLD program to society. No amount of analyses can provide a substitute.