Windows Defender and a dissertation on search algorithms

Dwight Silverman has a pair of interesting observations on the news that Microsoft Antispyware is about to become Windows Defender:

I mentioned above that there’s already an application dubbed Windows Defender. I found that by doing a Google search, which turned up many links to the existing package as the top results.

But if you do the same search at MSN Search, the top results are front-loaded with references to the Windows Defender renaming announcement by Jason Garms. In fact, the first reference to the existing Windows Defender product doesn’t show up until the seventh page of results at MSN.

Maybe Microsoft forced the results for its own entry higher on its search engine. Or maybe Google’s just slow to index blog postings. Or a little bit of both . . .

That first observation is interesting, indeed. Microsoft has an army of lawyers, and one would have to assume that no product naming decision gets publicly announced until there’s been a thorough trademark search. (At least the windowsdefender.com domain is owned by a guy in Seattle who is a contractor for Microsoft.) If someone made a public announcement like this without acquiring the trademark rights from the existing product, they were incredibly sloppy.

What about the search results? Is Microsoft really favoring itself?

When I looked at the MSN Search results, I found that a download link for the existing Windows Defender product was fourth on the list. (Hey, I’m even on that first page!) So it’s not like every reference to the existing product has been scrubbed.

I think there’s a (somewhat) more innocent explanation for the different search results for MSN Search versus Google. In my admittedly limited testing, I’ve seen clear evidence that the MSN algorithm emphasizes freshness much more than Google does. By contrast, Google’s algorithm emphasizes the number and quality of links to a given page (PageRank) and thus is inherently biased toward pages that are older and have had more time to acquire lots of links from high-traffic sites. So at least in this case it stands to reason that pages talking about the latest news on this phrase would rank higher at MSN Search than at Google.

For an example that isn’t Microsoft-related (and thus doesn’t have the possibility that Microsoft is unfairly favoring its own sites), try searching for Sony copy protection, a topic that has been much in the news lately.

Here’s the MSN Search results. Note that everything on the first page is about the current rootkit controversy.

Now try Googling the same words. Although there are lots of results about rootkits, I noted that the third item on the first page was a USA Today article from 2002. The sixth item is an undated article from KAOS2000 Magazine that talks about using marker pens to defeat Sony copy protection schemes used on a “new Celine Dion album” released in 2002. And the ninth link on the page is to a discussion at cdfreaks.com, also from 2002.

Those are interesting approaches. Knowing how those two search engines work can help me decide which one to use, but I don’t think either one is biased.

Scoble wrote a flurry of interesting posts on this some time ago. In this post, which I chose more or less at random, he says something I can wholeheartedly agree with: “Anyway, my point wasn’t to get into a rathole discussion on any one search term. It was to point out that at almost ANY search term you can find ways to improve the engine. But, I’ll keep hammering this one in until people get it and see that search is FAR from being done.”

7 thoughts on “Windows Defender and a dissertation on search algorithms

  1. Do the Google search again. The top result is news of the Microsoft Windows Defender product.

    People like Silverman try to maintain their journalistic integrity by using innuendo versus outright false accusations.

    Maybe someone should ask him if it’s true he beats his wife.

  2. Ablewasi,

    I know Dwight and he is one of the most scrupulous and honest journalists in the business. Your remarks are off base.

    (PS: The legendary question is actually “Have you stopped beating your wife?”)

  3. Ablewasi –

    Note that I phrased my speculation as a “maybe,” and I meant it in that way. I wasn’t even trying to state fact, just wondering aloud. Ed came along & provided some useful detail in response. That’s kind of what the blog conversation is all about, right?

  4. Re: phrasing as “maybe”. That’s exactly what I mean by innuendo versus an outright accusation. They are technically different, but the effect is very similar as any writer knows.

    The fact that you can have a “blog conversation” doesn’t excuse that kind of rhetorical device. It muddies the water we all drink from. It makes the blogosphere worse, not better and there are enough sloppily written blogs already.

    Rather than stating a “maybe” it would have been better if you had waited since you are aware that search engines take a little while to index recent information. If you had done that, you’d see that Google just needed time to index recent info and the Evil-Empire-conspiracy-theory-“maybe” wouldn’t have entered into a conversation where it doesn’t belong.

    It’s way too easy to take pot shots at Microsoft. If it’s not a question of journalistic integrity, then maybe it’s just an issue of journalistic laziness.

  5. Ablewasi, useless to attack Silverman in this way. You can have some doubts since Microsoft is not a saint, neither a devis, anyway.

    You can have doubts, Silverman can arise without killing anyone.

    The first thing I thougt after reding Bott’s article was “Ok, simply Microsoft bought the domain name from one of his contractors”. It’s a business practice, no scandals.

    The second was: “Weird, just on the 7th page”. Obviously, it’s so silly to think Microsoft moved the rank. So, Bott explained the different weights in the two algoritms.

    That’s all. I don’t see any laziness since Silverman doesn’t accuse any one but invites every one to take part in a discussion to enlighten some doubts.

  6. The majority of results from Google now relate to the announcemount. I think Ed’s hypothesis of MSN favouring freshness is probably more correct.

Comments are closed.