Search Engine Musings

Steve Balmer, not exactly the shy and retiring type, has been having a go at Google again. With a fine eye for a good headline he basically said that Google reads your email - while of course those nice people at Hotmail wouldn’t dream of such a thing.

Now regular readers of this blog (probably Mrs Trellis of North Wales) will know that I’m not exactly unconcerned about online privacy matters in general and Google’s desire to index everything everywhere in particular, but even I wouldn’t go quite that far.

Naturally no-one is sitting reading your mail - in this case specifically Gmail - but what they are doing is parsing through it to find subject matter to suggest adverts to display. How detailed that parsing is only Google know; and whether the CIA or whoever would be able to get personal information about you if they obtained access to the records is another open question. Since I wouldn’t use a Gmail account under any circumstances anyway it doesn’t worry me unless I’m communicating with someone who does.

From the search point of view the important thing to realise is that essentially Google isn’t really a search company - it’s an advertising company, and it’s increasingly THE advertising company. And it’ll use any data it can to push more, and more targeted, adverts at more and more people. The days when it was a simple search engine company with a simple interface and low revenue streams are long gone.

With considerable fanfare (Searchification?) we had the launch of the new style, new index, Live Search. Overwhelmed yet? Me neither, unless you count overwhelmed with crap search results. In what sounds increasingly like desperation in the face of Google’s continued growth, Microsoft claim to have 4 times as many pages in the new index. Looks to me as if they’re listing all of them before the ones that used to rank in their search queries!

To be honest I seldom bother to check MSN/Live now for my own sites - with so little market share they’re becoming an irrelevance - but I happened to be checking a client’s site which we rescued from a unindexed situation earlier in the year and which is now beginning to rank well in Google. I wanted an overall picture so I was doing a full check in all three engines. One of the first terms I checked was a very specific phrase that I’d used purely as a spider penetration test and it had quickly ranked in the top three on all three engines. I was astonished to find it had dropped out of the top 250 in MSN/Live. So I started to look at what was ranking for this phrase. Number 1 was a site which was pretty relevant for one of the words but didn’t appear to be at all relevant for the other, being in the wrong country entirely. The second word wasn’t in the body text, the title tag, or the description. By searching the source code I eventually found a single instance of the word - in a country list in a jump menu. Yet my client’s page is about the subject and has the phrase in both title and H1 tag.

I’ve looked at a few such searches since that one and they are no better at returning relevant results (with one notable exception - one of my own sites has good Google rankings but was lagging a bit in Live until now, although even there the results it did have have been turned upside down).

Maybe it’s still bedding in, maybe after a few more indexing runs it’ll improve. But didn’t they test it? Extensively. Didn’t they conceive that any poor relevance results initially would attract bad press and poor user response when they were making such a fanfare about it? They really need to do better than this if they expect to be taken seriously in the arena against Google. If they can’t then perhaps they should rename it Live and Let Die Search.