After months of high anticipation of what this "Google killer" would bring, Wolfram Alpha was finally opened to the public a week ago.

It's common practice in this day and age to compare any new public search engine to Google. Or more to the point, to debate whether or not the incumbent will be able to snag a piece of the giant's share of searches, currently just over 70% of the US market, but in most of Europe almost monopolizing at over 90%. This was the main point with Cuil, and it will be the main point with Microsoft's upcoming new web search engine, codenamed "Kumo".

Wolfram Alpha, of course, has to be different. And in this case, it actually is; so much so that it makes very little sense to compare it to Google. It's a "computational knowledge engine," which tries to "understand" your query (not so much in a natural language way, though it does try, but more as a mathematical equation) and then calculates the answer based on a corpus of "curated facts."

If you must compare it, look at Google's "reference tools," e.g. queries like "10 dollars in euros" (currently, about 7.26). Wolfram Alpha certainly does a better job at this (including nice graphs of the rate over the last year or last ten years). It will also, for instance, "compare York to New York" for you (and thankfully, it'll show you how it has parsed your request). Think of it as an excellent way to query sources like the CIA World Factbook or Wikipedia and perform comparisons.

Playing around with it, you'll also run into its limitations quite quickly. You can ask it the average temperature in San Fransisco or the average temperature in The Netherlands, but it won't compare the average temperatures in San Fransisco and The Netherlands. You can ask it to divide 12 by 3 and calculate the result as Dollars to Euros, and you can ask it for the  "currency of The Netherlands." But you can't ask it what "4 in US currency is in The Netherlands currency."

In short, while it allows for very complex mathematical queries, please don't ask it to check for more than two facts about facts at the same time. And if you're not looking for facts, you're completely out of luck: it can show you the color purple (the book), but you'll learn little more of substance than that it was first published in 1983 and won a Pulitzer.

It's tempting to digress on the roots of Wolfram Alpha and talk about Stephen Wolfram, his Mathematica software, and his book "A New Kind of Science", which attempts to answer, well, the question of life, the universe, and everything. But to be honest, this is probably as helpful as an explanation of Bayes' theorem is to understanding the usefulness of Autonomy IDOL. The main difference is probably that Stephen Wolfram, unlike Thomas Bayes, is alive and well, and his ego is an important driving factor behind Wolfram Alpha.

There's certainly a use for the engine, and you may want to bookmark the site for that reason. You may even want to use the API for this functionality once it becomes available. But "computational knowledge" has its limitations. Wolfram Alpha will, in fact, by using its principles of Mathematica and NKS, compute the answer to the question of life, the universe, and everything. The answer is 42. I'll leave it up to you to decide whether this just is the wrong question to ask, or if the answer is just really unhelpful.