Archive for October, 2005

Since take-out tastes pretty good out of those little cardboard containers, we’ve added our take-out options to the site. With the new My Web Badge, we’ll provide you with a customizable snippet of code that lets you publish your My...

Wondering who is behind the Boitho project? We have created a page with a little bit of info about every team member.

See http://www.boitho.com/people.htm

Wondering who is behind the Boitho project? We have created a page with a little bit of info about every team member.

See http://www.boitho.com/people.htm

Oilman beat me to the punch posting this one. Yesterday, Yusuf Medhi, MSN SVP, asked us to invite people to join the US MSN adCenter Pilot. I am going to extend the invitation to apply to all of our loyal blog readers. We want all of you to have a shot. See below for the invite and information on how to apply. 

     Brady Forrest, MSN Search PM

----
From: Yusuf Mehdi

As you know, we launched MSN adCenter --our next generation advertising platform -- in France and Singapore last month. This was an important first step to delivering our global vision to connect advertisers to consumers in a more meaningful way.

Now, as we prepare to launch the US pilot of MSN adCenter, we need your help in getting your friends and family to apply to participate in the self-service offering for small-to-medium businesses. This pilot represents a unique opportunity for small and medium businesses to be among the first to experience the more powerful tools and wide-ranging benefits of MSN adCenter. Our paid search offering will help advertisers:

• Learn by accessing comprehensive data to plan more strategic campaigns,
• Connect by using advanced demographics to target the right audience at the right place and time, and
• Refine by making meaningful changes on the fly with features for greater flexibility and control.

If you know a business-owner who would be interested in advertising their business on MSN Search, please forward this email to them and encourage them to apply to participate in the US Pilot by completing our online registration form at http://advertising.msn.com/adCenterPilot/89620.asp.

 Please note that entry in the US pilot is by invitation only. We will select participants on a rolling basis from those you refer and who express interest. During the US pilot, MSN adCenter will be delivering text-based advertisements on up to 25 percent of MSN Search traffic; the remaining traffic will continue to be served through our partnership with Yahoo.

I hope that you will support this important initiative by sharing this information with your friends and family. We look forward to serving their advertising needs and gathering their valuable feedback on our products and services.

 thanks,

 Yusuf

As we release more instant answers, we get a lot of great feedback on them.  One common question, also brought up by Scoble recently (with follow up), goes something like this: “I like it when you show the stock answer for queries like msft and intc, but why doesn’t it appear for other valid tickers like play?”

We could simply show the answer for all known financial symbols.  The problem is when someone searches for play, it could mean a lot of things besides the stock price of PortalPlayer.  A similar problem happens for searches on just a company name.  Before we show an instant answer, we try very hard to make sure it’s highly relevant.  Otherwise we would be pushing web results down the page for nothing.

Here’s another perspective.  When you search for msft, you would expect most of the web results to be about Microsoft.  What about searching for jobs, life or pets?  Would you expect most of the web results to be about 51job Inc, Lifeline Systems or PetMed Express?  If not, does it make sense for the stock prices of these companies to appear prominently?  We don’t think so, and Google and Yahoo both take a similar approach with these three tickers.  Also imagine a case where 17 instant answers can trigger for a specific query – does anyone want to see all 17 prominently on the page to keep each answer consistent?  However, if you add extra words like “stock quote” to the symbol or company name, the intent is clearer and we show the stock instant answer.

We realize that some people still just want the stock prices for jobs, life and pets.  There is also an element of serendipity when a good answer seems to appear out of the blue.  We are working on a few ideas for something that will handle these cases, without sacrificing relevance.

Do we always get this right?  Absolutely not!  If you come across any bad examples or have other thoughts on how we should handle this, please let us know.  We’ll take a look.

Jamie Buckley, MSN Search PM

Our friends over at www.ViaVirtualEarth.com are helping us run a contest for developers building Virtual Earth mashups.  If you have already written a cool Virtual Earth application or have a great idea for one, now is your chance to possibly turn that application into a prize of $1000.   It might be worth investigating the competition over in the VE Gallery like Poly9’s MSNBC news event map or Nikhil Kothari’s Photo Map. The contest ends October 14th, so check it out.

Viavirtualearth also has a bunch of interesting articles, tutorials, and examples to help you get started developing for Virtual Earth quickly. 
 

Alex Daley

MapPoint Technical Evangelist

Several people have asked us about compressing the XML we return from Yahoo! Search Web Services. We'd been reluctant to do it because some browsers and platforms claim to support compression without actually doing so. However, we can now return compressed data to clients that support compression while working well with systems that don't support it so well.

If you want the data we return to be compressed, do the following:

1) Use HTTP 1.1.
2) Pass in an 'Accept-Encoding: gzip' header.
3) Spoof your user-agent to be Mozilla/5.0 (or higher).

Don't forget to do the expansion on the other end!

Will this save you time? Quite possibly. XML compresses nicely, and savings of 50% on the data size are not uncommon. Depending on how busy your processor is, less data over the network may be fast enough to offset the cost of decompression.