If you build it, they won't care

All of the new TLDs offered a range of estimates of how many registrations they'd get, at 10%, 50%, and 90% confidence level. Their confidence was misplaced, since none of them have come close to their least optimistic 90% confidence level. The closest is .INFO which predicted between 3 and 11 million registrations and actually has 2.38M. The farthest is .PRO which predicted between 3 and 15 million and actually has 4900, but it's a special case due to being stuck in limbo for several years when its sponsor ran out of money.

You can fairly say that .INFO and .BIZ (with 1.28M registrations vs. predicted 3.8 to 30M) are successes, but pretty pale versions of success since each is in practice no more than a clone of .COM where people go if there's nothing good available in .COM. It's hard to think of a well known domain in either other than MTA.INFO, the New York City transit authority. Neither has come close to catching up with .ORG with 4.4M or .NET with 6.8M. The next biggest is .NAME where I counted 142,000 domains against a predicted 7 to 58M. Although my numbers may be a bit low since they sell individual names, it's unlikely they're anywhere close to recovering the $7.5M the proposal said they were investing. The three sponsored domains, as I've noted before, each have only 5% of the minimum they projected.

The comparison between .AERO and .TRAVEL is most interesting. The new .TRAVEL domain has an aggressive campaign to sell domains to members of trade organizations, and they say they already have over 10,000 registrations, twice what .AERO has, even though they haven't formally launched yet. A little spot checking showed an awful lot of .travel domains landing on a registrar's parking page, suggesting they're defensive trademark registrations, but we'll see soon enough.

In any event, it's clear that with the exception of .COM clones, selling new TLD domains to prospective registrants is somewhere between hard and impossible.

Domains are not directories

The .MUSEUM and .AERO domains both made attempts to build a directory structure into the domain. The .MUSEUM domain has a lot of geographic names and keywords you can string together to guess a name, like strong.museum.history.museum, and a wildcard record so if you guess wrong, they show you a web page with suggestions of names you might be looking for. It's not awful, but nobody I know is likely to type getty.museum into their browser address bar rather than getty museum into their Google search box.