Oon Yeoh

When I wrote about an interesting new "natural-language" search engine called Powerset in May, I ended my article by saying, "Perhaps, Google should just buy it up. Perhaps, that's just what Powerset wants."

Turns out I was wrong. It was Microsoft that's been eyeing Powerset and the Redmond giant bought it up last week.

For those of you who missed my column in May, Powerset contextualises search queries to improve results. Semantic search is supposed to be a potential Google killer. The rationale being that it can deliver more relevant results because it understands the content rather than just relying on keywords or the number or links to the site (which are the typical ways most search engines do it).

Microsoft's purchase of Powerset highlights its intense desire to play catch-up with Google and Yahoo! in the search business. In fact, it tried unsuccessfully to buy up Yahoo! itself, which would have placed it in second place.

Powerset won't increase Microsoft's market share because it's a new search engine that very few people have even heard of, much less used. So, what could Microsoft possibly have in mind for Powerset?

Usually, there are three reasons for acquiring a company: to acquire the company's user base, to acquire some unique technology, and to acquire the company's workforce.

In the case of Powerset, it has no substantial user base. It has interesting technology but nothing that Microsoft couldn't build up if it wanted to. The key reason is probably to acquire the senior staff at Powerset, some of whom come from the legendary Palo Alto Research Center as well as a former web architect from Yahoo!.

As mentioned in my May article, right now Powerset only provides a search and discovery experience based on content found in Wikipedia. This is incredibly limiting but it's amazing how good it is. Especially when it comes to natural language questions.

I'll give you some real examples. Try typing "Who was the father of Malaysia?"

In Powerset, the very first answer you get is "Tunku Abdul Rahman". In Google, the first result is a Guardian article entitled "Father of Malaysia savages Bush and Blair".

Okay, next question: "When did Malaysia gain independence?"

In Powerset, the first answer is an article entitled "Hari Merdeka". In Google, the first result is "The History of Malaysia".

So, Google's answer is not completely wrong but Powerset's is far more accurate.

I wouldn't go as far as to say Powerset is better than Google overall. When it comes to general search, I'd still use Google any day. But Powerset can be good when you need to find something in the form of a question, and especially if the information is something that can be found in Wikipedia.

Why not just use Wikipedia's built-in search engine, you might ask. The answer is that Wikipedia's search engine is lousy - so much so that in the past, before Powerset came along, whenever I had to find something in Wikipedia, I would use Google to locate it.

Powerset's main disadvantage right now is that its search results are limited to content on Wikipedia, which is but a tiny subset of the Internet.

However, the systems it has developed is generic, so any other site could also be enhanced by Powerset. It could theoretically be scaled up to search the entire web but that's no mean task. Applying Powerset across billions of web pages is a different kettle of fish from using it to search Wikipedia.

However, it now has an owner with incredibly deep pockets. Powerset was reportedly bought for US$100 million.

Thus, Microsoft could easily devote the resources needed for Powerset to scan, read and understand literally the World Wide Web.

Could it be a Google killer? It's too early to tell. The potential is there but for now, it's just potential. What's to stop Google from investing in its own semantic-based search technology if it feels threatened by this approach to search?

There's another uphill battle for Microsoft/Powerset: people are creatures of habit. All of you who are reading this article right now probably use Google as your default search engine. Google may not be perfect but it's pretty good. Unless something comes along that is disruptively better - and it takes a lot to be disruptive - people simply aren't going to switch.