Slashdot Mirror
Microsoft Lost Search War By Ignoring the Long Tail
Art3x writes "When developing search engine technology, Microsoft focused on returning good results for popular queries but ignored the minor ones. 'It turned out the long tail was much more important,' said Bing's Yusuf Mehdi. 'One-third of queries that show up on Bing, it's the first time we've ever seen that query.' Yet the long tail is what makes most of Google's money. Microsoft is so far behind now that they won't crush Google, but they hope to live side by side, with Bing specializing in transactions like plane tickets, said Bing Director Stefan Weitz."
Company releases an inferior product, much later to the game than competition, makes excuses for failure, water still wet.
"we've got trenchcoats and bad attitudes" - John Constantine, HellBlazer
.
Search engines are all about people looking to find stuff. A good portion of what people look for are probably new things that are happening now.
So, Microsoft goes off and designs a brand new "bet the ranch" search engine, without even knowing how its customers use such a service. Yes, that sounds like Microsoft.
Someone should tell Medhi that it also helps when you don't game the search results to fit your corporate agenda.
From time to time, I try out the following query on Bing: "Why is Windows so expensive?"
The day that the first result returned is NOT a site about Macs being expensive is the day I'll start to take Bing seriously. Until then, I'm sticking with Google, which is at least honest enough to properly index anti-Google queries.
Microsoft is so far behind now that they won't crush Google, but they hope to live side by side...
The same way the Zune lives side by side with the iPod.
That's our life, the big wheel of shit. - The Fat Man, Blue Tango Salvage
I would say they lost by:
1. Being too late. Search engines have been around for many years. You can't easily launch a search engine now without a massively improved user experience over what is already available.
2. Not being trusted, I don't want to use Microsoft's search engine as it may subvert the results to promote their wares.
3. Stupid name. Every time I hear "Bing" I think of Ned Ryserson from the film Groundhog Day.
4. OTT interface, I don't need a big background when I'm looking for stuff.
http://www.joelonsoftware.com/articles/ThePerilsofJavaSchools.html
Perhaps Microsoft just cannot think like that? To be clear, Microsoft saying that maybe Google and Bing can perhaps exist side-by-side is a clear admission of defeat. Microsoft never says that, so you know the situation is bad. I just can't understand why they got a bee in their bonnet and wanted to chase Google in the way that they have. It was clearly a knee-jerk thing and they hadn't clearly thought about it. The only major difference they did was change the name from the stale MSN Search name to something they thought was cooler - Bing. Nothing else changed.
To not take into account that people search for many random and obscure things put together that won't have been recorded before (language is a very broad thing and what people search for is also time-based i.e. NOW), and not to have some sort of logic to aid with that, is utterly unforgiveable. What the hell are Microsoft Research doing?
I don't think Bing will ever out-Google Google. So it's strange that they don't try to identify problems with Google and address them. They seem to start out with the assumption that Google is perfect, so the best path forward is to do everything just like Google, only more so.
The big problem with Google is privacy. Why not try to make a search engine that doesn't track what you do? I'd pay a subscription for such a thing. Maybe most people wouldn't, but I would. Search is such a big market that 5% of it is still huge. Maybe 5% of the people in the US would pay for private searching.
MS has had a kind of bullying culture for a long time, and they've declared war on open source, so we've viewed them as the bad guys for a long time. But windows is a heck of a lot more open than the iPad, and their business model isn't based on data mining. In a lot of ways, they've been left behind by many of the most toxic trends in the industry. They should listen to some of the things that we linux folks have been saying, and try to fit them into their pitch when they can. Talk about the value of controlling your own data, of privacy, of letting anyone who wants to write a program and distribute it, of being able to install your software on whatever hardware you want. That's not snake oil -- it's good stuff.
The strange thing is that they've missed those toxic trends not because they value the good alternatives, but because they're big and sluggish and not very agile. They've just been left behind. And all they want is to catch up so they can turn the same screws on us that Apple and Google turn. It doesn't occur to them to make the kinds of arguments I'm proposing here.
Why not try to make a search engine that doesn't track what you do? I'd pay a subscription for such a thing.
How would they keep track of who has subscribed if they're not tracking people?
Performance is comparable to Google, but I get the impression that Bing's results are more skewed for people looking to purchase things on line. Try some basic 'how to' queries (how to caulk a window, how to make pancakes, etc) and see if you get more product related hits returned near the top with Bing than with Google.
This isn't a criticism, just an observation. It could be a smart thing for MS. It will help them squeeze more advertising dollars out of smaller market share.
It seems to me that Bing may be a better tool for shopping than Google is, but Google is a better tool for searching than Bing is.
Bing's problem then becomes that there are several better tools for shopping and comparing prices than Bing offers.
Remember Cuil? They were originally talking about the "long tail"; they wanted to have a bigger index than Google. Cuil is mostly ex-Google people, and they thought they could re-do Google at lower cost.
Didn't help Cuil.
There's ongoing effort in search engine development. Unless you pay close attention, though, it's invisible. A few years ago, around 2007, Yahoo introduced about fifty specialized search sub-engines. These understood weather, stocks, sports, celebrities, movies, and similar popular search topics. They focused on areas that have a strong structure, and need a lookup engine that understands that structure. For about six months, Yahoo was way ahead of Google on such searches.
Didn't help Yahoo. Google implemented something similar and caught up. Now everybody does that.
It's not clear that the Twitter search is a win. Bing announced they were going to do Twitter and Facebook searches, and a day later, Google announced they'd do that too. Google implemented Twitter search, and apparently Bing didn't. Twitter search just seems to clutter up Google results.
In the last year, Google has become much more aggressive about interpreting queries. Google tries hard to infer from the query words what the user is really looking for. This tends to work for popular queries (since it's based on statistics from other queries) and doesn't work too well for unusual queries. For hard queries, you need to use explicit operators ('+' and '"') with Google more than you did a year ago.
The big search engines are still doing badly at de-rating sites which are basically link farms. When you're searching for a product, and you get a hit that's just some site with ad links to other sites, that's a fail. Search for auto parts, and you're likely to get "parts.com", "thepartsbin.com" and "who-sells-it.com", which are just "portals". They don't even return pages that are actually about the part in question. ("thepartsbin.com" pages are all essentially the same, except for keywords inserted for SEO purposes.) Search engines need to look at the business behind the web site. If a business has a million commercial-looking web pages, and a total business volume of a few million dollars, they're probably bogus. That's a part of the "long tail" you don't need to visit.