UpNext and other map-based systems offer one significant advantage over text-based interfaces: geospatial search. When people are looking for something to do, they usually know the general area where they want to go before-hand, but don't know exactly what they're looking for. A keyword based system is only efficient when you already know what you want to do. It doesn't help the process of discovery.
The second advantage of UpNext is an improvement over current 2D search offerings, which are usually based on Google maps, Yahoo maps, or MapQuest. While most people focus on the 3D, the other enhancement of UpNext is real-time rendering. Other map-based systems use static images or photos; they're unable to change on-the-fly to better highlight information. Our spotlight feature showcases the advantages of real-time rendering in search: buildings of interest glow, and others become transparent. In the future we plan to have the map do a number of interesting things in real-time to give people a feel for the neighborhood they're in.
Then there is 3D, the topic of the comments above. Yes, 3D doesn't convey mounds of information, but if you're doing real-time rendering, you might as well do it in 3D. Beyond looking compelling, the perspective information does convey a feel for the area you're looking at, and we hope to further improve this in the future to attain a stylistic impression of each area of Manhattan. That 3D contains occlusion is easily mitigated by simply rotating to a birds-eye view.
Finally, that there is no standard for 3D navigation is simply because the technology is only recently breaking out of the gamer universe. Regardless, if everyone waited for standards before innovating, the internet would be in a very sorry state.
Thanks for the comments! Raj Advani, UpNext