Doing a good job of helping Search users to correct misspelled queries is super important for two main reasons: a) 5 billion crawled docs and a bleeding edge ranking algorithm can’t do much if the query isn’t spelled right and b) more than 10% of all searches are misspelled! So we made sure our new search engine included a revamped spelling correction system that’s much better than our old one. To improve the speller we worked...
Read More

We’ve found quite a few people have added on extensions to the beta.  It’s great to see more people playing with the new engine. Here’s a tool from DEVPPL that can be used to evaluate your rank on the Beta.  Martin created an MSN Search Beta desktop widget. And Ryan has created a Firefox plug-in for the Beta. Thanks for your contributions!    osh  
Read More

First of all, thanks for your valuable feedback via the website and around the blogosphere.  I read the feedback coming in from the site as a part of my job. Thankfully we’re past the performance hiccup of day 1 – the initial beta instability made for lots of uninteresting feedback!  On the feature request side, we’re seeing lots of interesting ideas, some new and some old. Our advanced user base wants a more...
Read More

We launched the beta just over a week ago and there’s been a lot of discussion about what we’ve been doing. Reviews of the beta: WebProNews notes that the beta feedback has been mixed.Mike Hall loves the fact that we can help him with his algebra homework.Angsuman Chakraborty thinks that we released the Beta too early. He also questions the veracity of our site descriptions. I want to respond to this by re-iterating that there is no...
Read More

Now that we have a beta, people are starting to pay attention to whether their sites are in our index.  The two most common questions we get are (1) why did you not crawl my site, and (2) you crawled page X, but its not  in your index why?  Let’s take these one at a time. Why did MSNBot not crawl my site?  The answer to this is not straightforward so I will mention a couple of things that are worth  considering. The...
Read More

Taking a cue from IE team blog and from Robert Scoble’s Corporate Weblog Manifesto, allow me to lay out some basic principles and aspirations for the MSN Search blog:  Honest and Respectful.  We will tell the truth and stay the high road.  ‘nuff said. Useful.  We will announce new products, features and releases related to MSN Search.  We will answer frequently asked questions, make clarifications, and once in...
Read More

As you may have heard, some screenshots from an internal pre-release build of MSN’s desktop search product were recently posted on the Neowin site.  While we’re flattered by the attention and interest, this was an unauthorized disclosure.  As many of you know, we have been working on a PC search product for some time, and have demonstrated some early prototypes over the past several months.  We are still on track to...
Read More

As mentioned in a previous post on this blog, we recently received our first few MSN Bombs.  One of these was the phrase “more evil than satan”, which brought up both Google and Microsoft in the top 10 results.  A number of you sent us feedback on this.Well, we just pushed out our latest round of automated relevance improvements and they changed the results. Now neither Google nor Microsoft rank in the top 10 any more for...
Read More

As you can imagine it’s been busy here. In the 2 days, we’ve served millions of page views and queries, rolled out a couple of system updates (note the improved performance) and yet I was still able to go see The Incredibles last night (it was great!). We’ve had over 1000 pieces of feedback sent to us, in addition to what we’ve seen in WebmasterWorld, the Search Engine Watch Forums, our newsgroup, and on blogs. The top...
Read More

Several people have asked how the system performs a “Near Me” search for food bank near Seattle, WA so I thought it would be a good idea to explain how this is done here on the blog. As web pages are rolled into the index, the code looks for strings of text that are location references.  In particular, text that is formatted as an address is given a high degree of confidence.  When an address is found, tokens are added...
Read More