I'm enjoying the last week of summer, so in lieu of a new post, I'm reposting a 'best of' entry this week. This was originally posted May 13th, 2006.

NOTE: There is a political refernece below, as it is a topical situation that got me thinking about trust communities in search.  This blog is a-political, and the scenario is used as it is the one that sparked the idea. I take no stance on whether the claims made by Mr. Snow are valid/invalid.

I was reading some news sites this week, and was reminded that this was the first week of US President George Bush's new press secretary, Tony Snow. 

Before he gave his first press conference, he did something interesting.  He sent out press releases questioning the validity of comments made by the New York Times, USA Today, and other publications.

So this got me thinking.  Playing devils advocate, suppose that he's right. If I trust Tony Snow (based on his historical record of trust worthiness), I may now discount results from these media outlets in favor of others.  But for me to discount these sources when searching, I can't. Even if I cease to trust them (or trust them less), they show up in the the rankings per Google or Microsofts opinion of their relevance.

The search engines from Google, MSN, and Yahoo have their own algorithms to consider relevancy. One of the things these search engines do provide is a level of filitering for “safe content“, blocking out material that may be considered objectional (i.e. these block pornography results). 

What they don't do is consider in the rankings is the levels of trust of an individual or of community. What I want to see is something that goes to the next level, don't just block what's objectionable, show me the results that are relevant to me based on trust.

Perhaps one of the media outlets Mr. Snow referenced, let's pick a fictional name, say MakeBelieveReporting, Inc., is regularly mis-reporting information or is slanting stories towards a particular political viewpoint.  I may cease to trust that organization to provide news to me, and would like to rank them lower in my personal results when searching for news, if not remove them altogether.

When I search for news, perhaps there are certain stations / periodicals I trust - for example the Wall Street Journal, the Financial Times, CNBC, and my friends John Smiths blog. These are entities - regardless of web site traffic or the opinion of the search engine I'm using -that *I* trust to be accurate and provide me information.  I do not, however, want results from news outlets that are part of MakeBelieveReporting, Inc. as I have ceased to trust them.

What I'm thinking of isn't based on assumed trustworthiness based on click traffic, this is based on trust relationships.  Even if I visit a site twice per year, it could be far more relevant to me than a site that is viewed more regularly by others.

And my community of trustworthy providers could be extended based on the feedback of those people I trust. 

There's the concept that if person A trusts person B, and person B trusts third party C, that person A shoud likely trust Corporation C based on the fact that he trusted person B's judgement.

In the previous example, I trust my friend John Smith who writes a blog.  If John trusts the Crosby Herald, and I trust John, then I too could trust the Crosby Herald and have it included in my community of trust that is reflected in my search results.

Think about the days before Axciom, TRW, and credit reports. People vouched for other people to get jobs, apartments, loans, etc.

When you sign for a loan and you are not a known entity, you need a co-signer or guarantor. The bank says, I don't know if I can trust this person, but I trust the co-signer.  The co-signer also trusts the loan recipient to pay the money back.

If someone co-signs for a loan for me and I decide not to pay it, there are financial responsibilities that are then taken on by the co-signer. The co-signer will trust the recipient less, as a result of mis-placed trust, the bank may stop trusting the co-signer's ability to identify a trustworthy loan recipient.

In another example, suppose you make plans to go out to dinner with your spouse on Friday night, and when you ask her where they'd like to go, she says “you pick - I trust you.“  If you're new to the area, you may ask a colleague - whom you trust - for a recommendation of a local restaurant. If you go to the recommended restuarant and you end up getting food poisoning from the meal, you probably will not look to your colleague for advice on restaurants in the future - and you - who vouched for the restaurant -will likely end up at a restaurant of your spouse's choosing next time around.

Your spouse trusted you, you trusted the colleague and when the information relayed turned out to be bad, two things happen. You cease to trust the advice of the colleague (atleast in the context of cousine), and your spouse trusts you less as the broker of the information.

Once you start adding in trust, you also need to be able to trust in context.  That same colleague from work may not be someone I trust on picking restaurants, but may be someone I look to as a source on technology subjects.

What we need is search that includes both consideration of these communities of trust, where we as participants in the web determine who is trusted and who is not, and provide the ability to apply trust in context.

By introducing contextual trust as a first-class citizen in search, it has the opportunity to both provide results more relevant *to me*, and as trust=traffic=revenue, provides a financial incentive for providers to be trustworthy.

That's my two cents  - what do you think?

Chapter Three of my upcoming book focuses on the work being done with information cards and in the identity metasystem by people outside of Microsoft. The chapter covers third parties and open source projects,  focusing primarily on the folks building identity selectors and security token servers.  In the process of researching that chapter, I ,of course, ran across the work of Chuck Mortimore. If not famliar with his work, Chuck has built out a Java Relying Party, an identity selector plug-in for FireFox, and his site ( http://www.xmldap.org) issues managed cards.  His identity selector has even been enhanced to handle interop with OpenID (see screenshot below).

Needless to say, I was impressed with his work, and reached out to him about including screenshots of his work in that chapter. He was very gracious and gave his approval. As I was wrapping up the book, one of the readers of this blog asked if we were going to have support for Java in the book. Initially, for relying parties, I'd only committed to the publisher for ASP.NET and PHP. In the pre-.NET world, I actually was an early adopter of Java  (heck, I even hired Gary Cornell, of Core Java fame, to come to Boston and train my team on Java), so I thought what the hell, and decided to  have a go at it. As I was dusting off my core-java books to write the sample, I thought to myself, if I was a java guy, who would I want a sample from? A Microsoft guy who hasn't written any Java code in awhile? Probably not

I thought of who - if I was a reader - I'd like to see the Java sample come from. A big fan of his work over at xmldap.org, I reached out to Chuck and asked if he'd be interested in contributing a java sample for the chapter. I am really pleased to announce that not only did he agree, he's already sent me the code. If you've not done so already, definately check out his site, he's doing some great work.

There's a blogger from Australia who had run across the DotNetRocks podcast, and had some questions about what InfoCenter is and what InfoCenter isn't.   Quite frankly, I've been amazed at the response, considering the CTP isn't yet evailable.

At any rate, one of the questions that came up was around the value and/or role of the aggregator of feeds  - not just software, but the humans who inject domain expertise and experience in putting together an aggregated blog. People who involve software and people to bring together a noise-free, targeted feed.

For this, I look to the last letter in RSS? That S stands for Syndication.  Effectively, the way I see the world is that every blogger is a production company, generating syndicated content. While we can broadcast our own content via our blogs, there's also interest in repackaging our content and leveraging it in other places.

Just like in television, people generate revenue by hiring program directors / editors that compile a group of syndicated content into a product or programming schedule and sell ads based on viewership.

While the underlying value is in the content, there's tremendous value in the aggregation of particular content by people we trust and who we hold up as experts in a particular area.

We all have the opportunity to become program directors of RSS content, what needs to be figured out is how the revenue stream works back for the creator of the content.  In the syndicated television world, you buy rights to a show for a particular period, for a particular market. The payment, as I understand it, is established up front.

On the web, it would seem this would be more fluid, and as a result there needs to be a way for either the syndicator to specify ads to be shown in conjunction with their content, or for networks/program directors to provide incremental revenue back to the networks/program directors.

NOTE: There is a political refernece below, as it is a topical situation that got me thinking about trust communities in search.  This blog is a-political, and the scenario is used as it is the one that sparked the idea. I take no stance on whether the claims made by Mr. Snow are valid/invalid.

I was reading some news sites this week, and was reminded that this was the first week of US President George Bush's new press secretary, Tony Snow. 

Before he gave his first press conference, he did something interesting.  He sent out press releases questioning the validity of comments made by the New York Times, USA Today, and other publications.

So this got me thinking.  Playing devils advocate, suppose that he's right. If I trust Tony Snow (based on his historical record of trust worthiness), I may now discount results from these media outlets in favor of others.  But for me to discount these sources when searching, I can't. Even if I cease to trust them (or trust them less), they show up in the the rankings per Google or Microsofts opinion of their relevance.

The search engines from Google, MSN, and Yahoo have their own algorithms to consider relevancy. One of the things these search engines do provide is a level of filitering for “safe content“, blocking out material that may be considered objectional (i.e. these block pornography results). 

What they don't do is consider in the rankings is the levels of trust of an individual or of community. What I want to see is something that goes to the next level, don't just block what's objectionable, show me the results that are relevant to me based on trust.

Perhaps one of the media outlets Mr. Snow referenced, let's pick a fictional name, say MakeBelieveReporting, Inc., is regularly mis-reporting information or is slanting stories towards a particular political viewpoint.  I may cease to trust that organization to provide news to me, and would like to rank them lower in my personal results when searching for news, if not remove them altogether.

When I search for news, perhaps there are certain stations / periodicals I trust - for example the Wall Street Journal, the Financial Times, CNBC, and my friends John Smiths blog. These are entities - regardless of web site traffic or the opinion of the search engine I'm using -that *I* trust to be accurate and provide me information.  I do not, however, want results from news outlets that are part of MakeBelieveReporting, Inc. as I have ceased to trust them.

What I'm thinking of isn't based on assumed trustworthiness based on click traffic, this is based on trust relationships.  Even if I visit a site twice per year, it could be far more relevant to me than a site that is viewed more regularly by others.

And my community of trustworthy providers could be extended based on the feedback of those people I trust. 

There's the concept that if person A trusts person B, and person B trusts third party C, that person A shoud likely trust Corporation C based on the fact that he trusted person B's judgement.

In the previous example, I trust my friend John Smith who writes a blog.  If John trusts the Crosby Herald, and I trust John, then I too could trust the Crosby Herald and have it included in my community of trust that is reflected in my search results.

Think about the days before Axciom, TRW, and credit reports. People vouched for other people to get jobs, apartments, loans, etc.

When you sign for a loan and you are not a known entity, you need a co-signer or guarantor. The bank says, I don't know if I can trust this person, but I trust the co-signer.  The co-signer also trusts the loan recipient to pay the money back.

If someone co-signs for a loan for me and I decide not to pay it, there are financial responsibilities that are then taken on by the co-signer. The co-signer will trust the recipient less, as a result of mis-placed trust, the bank may stop trusting the co-signer's ability to identify a trustworthy loan recipient.

In another example, suppose you make plans to go out to dinner with your spouse on Friday night, and when you ask her where they'd like to go, she says “you pick - I trust you.“  If you're new to the area, you may ask a colleague - whom you trust - for a recommendation of a local restaurant. If you go to the recommended restuarant and you end up getting food poisoning from the meal, you probably will not look to your colleague for advice on restaurants in the future - and you - who vouched for the restaurant -will likely end up at a restaurant of your spouse's choosing next time around.

Your spouse trusted you, you trusted the colleague and when the information relayed turned out to be bad, two things happen. You cease to trust the advice of the colleague (atleast in the context of cousine), and your spouse trusts you less as the broker of the information.

Once you start adding in trust, you also need to be able to trust in context.  That same colleague from work may not be someone I trust on picking restaurants, but may be someone I look to as a source on technology subjects.

What we need is search that includes both consideration of these communities of trust, where we as participants in the web determine who is trusted and who is not, and provide the ability to apply trust in context.

By introducing contextual trust as a first-class citizen in search, it has the opportunity to both provide results more relevant *to me*, and as trust=traffic=revenue, provides a financial incentive for providers to be trustworthy.

That's my two cents  - what do you think?