Is friendfeed enough of a ghost for you?

I certainly do at least a couple dozen searches on Google or my blog a day “Victus spiritus economy of minds” etc.

I get to enjoy shifts in perspective as well. Early hunches transform into more polished perspectives.

I'm interested because of the gears turning in my ratchedy brain that dream of a complete personal user profile that iteratively improves based on social media usage, blogging, purchasing habits, social graph, beliefs, passion…

I basically want a machine readable version of my web pursuits, a digital ghost, and I'm willing to create a business to do just that.

Side note: have you considered passively aggregating users social data to “automatically” answer some of the hunch questions. I'm starting with a short tag memory derived from semantic extraction, but have plans to build up a better data profile on users in the future by slowly filling in a machine based image of their likes/dislikes. The long term goal being creating personalized virtual assistants to aid in search.

The algorithm itself defines the behaviour of the people that create the content. Every marketer focuses on SEO, and these content farms are no different. They aren't gaming anything. They are simply behaving in the way that the algorithm(s) value.

As you point out Chris, the incentives are primarily economic, so the behaviour is to create content that, on average, will drive increased economic value to someone like demandmedia.

A question we should ask ourselves is whether or not these behaviours are the right ones to incentivise?

Should we be looking at other incentives that emphasise different positive behaviours -> socially, culturally, creatively or even personally?

Should every page rank consider the social reputation of the people posting the link? How do you measure social reputation?

Can we create algorithms that do this? Or is it always a double edged sword? i.e., do you always enable both positive and negative behaviours no matter what?

P.S. I'm glad companies like demand media are around. If it wasn't for them, we wouldn't have to worry or innovate around this stuff for many more years. The more low quality content they create, the greater the demand for the algorithms to change and new solutions like hunch to be adopted. If Google open sourced their page rank algorithms… with visibility, these negative behaviours would have happened sooner and they would have been addressed sooner.

Sure stuffing happens on site (can get sites sandboxed) but that is also a occurrence within the social conversation as well.

So it all boils down to preference in search platform for finding information for the user.

Given a choice I think most would rather have links/products recommended to them by people in their network as this comes with an extra layer of security. For example, If I buy a digital camera off a friends rec, I have someone to call if something goes wrong. If I buy it because it was on the first page of Google and it turns out to be crap, I don't think I'd get a response from Larry Page if I sent him an email.

When it comes down to it Social Recommendations are the most powerful but they will always be limited by your individual network, everyone will have holes.

This is why I think the most interesting company to come out in this space is Vark. They go after your 2nd 3rd degree connections, exactly the people you aren't aware of but are in a close enough social sphere.

What percentage of searchers know a great copyright lawyer? Or have a friend that publishes quality content on the web? 'Search within your social graph' plays great with us that influence and report on the web, but I've gotta imagine we're in the top .001% of users.

True, but part of Google's strength comes from the perception that they rank and filter all of the content on the web. From a Brand perspective I'd think they'd want to index this content even if it wasn't profitable, if only to maintain this illusion of comprehensiveness.

But the future of search is not reduction, since:
Meaning = Data augmentation with [my] data
Learning = Self Organization of Data (which build context, which leads to Information, which leads to knowledge …)
In other words we can teach a machine to understand, thereby reducing links to pointers of addresses only and eliminating them as an information organizational crutch.

In other words the filter works by expanding and then definiteness. The expansion is out of the controll of content provider, very different from map reduce or any crutch algorithm currently used. The typical spam content will just fall out since it doesn't provide anything new/useful when put into context.
Which will turn Google's intent upside down, since it can be extremely precise for a machine to find what you look for. Google needs fuzziness to grow.
Which brings us back to the Demographics, who has the best demographics? Social networks.
In other words the future is a combination of monetizing the new demographics with search.

yes, am talking about who owns the data but also user expectations of the data they'll find at each destination.

So I think as more data streams get added to our social networks (so beyond “what are you doing?” and into “where are you”, “who are you with”, “what do you think about X” [which btw is where I think hunch leads people], etc.)…the more that gets added, the more likely you are going to see solid niche data go through social network streams…

I also think the subtle point lost here is very much what people initially didn't 'get' (myself included) about long form blogging…it's not always about sharing the data with other people, sometimes it's just about sharing the data with a future version of yourself (I can't tell you how many times I refer back to my own blog for 'how' or 'why' I did some bit of code…and more and more I find myself tweeting interesting links just as a casual way for me to bookmark something and be able to forget about it until a future version of me has time to go back and digest the actual content)…

That's the million dollar question. At the moment, the answer is whoever has ultimate access to the DB on which it's stored.

But I see that changing over time…

And that doesn't even include potential innovation: if Twitter or Facebook search parsed on relevance and source proximity over time you could see it being a highly valuable machine for plane tickets, dishwashers, etc.

Here is my take on social search math:

_x_ Size of social graph (what is average?)

__% of graph accessible on given platform

__% of people with familiarity of a topic (some)
__% of people with expertise on a topic (few)

__% of people who care enough to review topic

__% of people who have already reviewed, or
__% of people who can review quickly for you

_?_ ability to triangulate all of the opinions

Fred's math <> Chris' math <> my math <> my Mom's math etc.

IMO, if we had to rely on social search for most topics, we'd spend a lot of time waiting for results that were inconclusive. Ideally, I'd like to see a blend of hard data, expert reviews, user reviews & social search.

these systems can be searched

That's the fundamental difference between passive laid-back social discovery and active (I want it NOW!) search.

But I think there's a lot of scope for leveraging passive SN shares into real-time search results (even though it does come with quite a lot of semantic baggage).