There are three things at work here:
1. How stale the index/data is.
2. The coverage of the data.
3. The processing algorithms that do magic on the data to return results.
These are the key things every search engine deals with.
Google, Yahoo!, and MS Live compensate for stale data by scheduling their crawlers very cleverly. Google has published a few blog posts a while back about how it does it (I'm sure it's even more improved now).
The breakthrough of W|A is number (3) in my list above. They kick ass and frankly make search much more useful than anything we have now. You're no longer searching for pages that have the info, but getting answers directly. A Cambridge, UK, startup called True Knowledge does this too, and it's impressive (W|A is much better in data presenation though).
So for an alpha product, please do not judge it by have stale or incomplete data. It simply can't have all the data fresh from the get-go. Judge though by how good or bad are the algos that power the search. The data is relatively easy to get hold of when compared to writing awesome analysis code.
Even with academic queries it's very hit or miss. When you get something that returns relevant results it's a Eureka! moment, and you'll be impressed. But then you try a similar query and get zilch.
I suspect academics will try their queries on Wolfram first and turn to Google if that doesn't work. But after seeing too many "no results found" messages, they might just skip the Wolfram step entirely. Google might not have the fancy graphs, but it almost always returns some kind of relevant data.
Honest curiousity: What computational-biology question would you ask Wolfram|Alpha that you would expect a better answer to, vs. either Google, conventional journal search, or asking your own local computing resources? (In particular the questions that I know enough to ask, such as they are, would seem to be more suited to local resources.)
(In fact, I'd open that question up to any other similar discipline.)
It looks like the launch of this is going better then Cuil and most people seem to understand that they should give it a chance(time) to mature into something extremely useful! It looks very promising!
Google began in 1997 or 98. I heard of it in 2000 but did not start using it til 2002/2003.
I am quite interested what is the business model for Wolfram Alpha. It seems that is rather focused on data analysis then searching.
It might be tough to rely solely on ads, but perhaps companies may be willing to pay to run their computational engine against their own data. It could be a great tool for exploring data.
The queries that we see right now could be hard to monetize. Lack of business cases and probably impractical to go mainstream.
Right now their are not any serious competition for Google. Refining the engine might help a bit, but I don't think their are going to replace ordinary search engine.
Of course it's probably not going to return anything relevant when you enter your own name, why should it? It's certainly not Google. I get the feeling the author doesn't understand that.
Also, they have only provisioned for like 2000 requests a second [1], which seems pitiful for launch day when every one and their mother will hit the site.
Can I refresh the page to see the distance between the Moon and New York change in real time, including accounting for the 1.3x10^-9 Earth-Moon recession speed? If so that's pretty cool!
I'm going to throw out there that Wolfram|Alpha might scrub it's sources' data more thoroughly. How can you trust unemployment data from a month ago? 2006 may have been some major study by INSERT_MAJOR_ORGANIZATION_HERE.
These are the key things every search engine deals with.
Google, Yahoo!, and MS Live compensate for stale data by scheduling their crawlers very cleverly. Google has published a few blog posts a while back about how it does it (I'm sure it's even more improved now).
The breakthrough of W|A is number (3) in my list above. They kick ass and frankly make search much more useful than anything we have now. You're no longer searching for pages that have the info, but getting answers directly. A Cambridge, UK, startup called True Knowledge does this too, and it's impressive (W|A is much better in data presenation though).
So for an alpha product, please do not judge it by have stale or incomplete data. It simply can't have all the data fresh from the get-go. Judge though by how good or bad are the algos that power the search. The data is relatively easy to get hold of when compared to writing awesome analysis code.