That is the tough issue here. The hype has been immense.
All Wolfram has to do is to allow developers access to the API. I'm just not totally sure how he will allow access to the program right now.
Google's search engine isn't incredibly useful to the person who doesn't like using it. But Alpha has the possibility possibility of being useful to people who wouldn't even visit the site.
At the very minimum, it has a lot of potential for search aggregation. Let's say that someone wanted to create a site that aggregated travel information. It sounds like a developer could basically create mini-travel guides on demand for specific regions or towns. So you would quickly get info on crime, cost, food, weather, and tourist attractions. Or you could actually just rank travel places based on certain criteria. So if I wanted to find a cheap place that currently had warm weather, beaches, and low crime, I might be able to get a lot of useful information. I wouldn't need to use Alpha myself, somebody would have just had to develop a site to aggregate the results.
I have no idea what it's killer use will be, but this has the potential to be very useful to developers. So once developers start taking advantage of it, I believe it would be successful.
Cuil results really were nonsense, and still are today. It wasn't some over-hyped bad publicity that was Cuil's downfall -it was their crappy software.
If WA is good, people will use it, and the talking heads won't have any say in the matter.