5 Comments »
Social search is a type of web search that has been touted as the basis of the search engine of the future. Many have attempted to tackle the challenge of social search but so far all have failed. Back in March of last year, Will M of AllFacebook interviewed Aditya Agarwal, Director of Search Engineering at Facebook, about the future of Facebook’s social search. Aditya hit on some of the core competitive advantages that Facebook has over other social search engines, the primary advantage being Facebook’s accurate mapping of the social graph.
The phrase “social search” is somewhat of an oxymoron though. Google’s search algorithm is based on inherently social features. Web publishers link to other publishers as a form of reference and those links are counted as votes that help determine a page’s “Page Rank”.
This linking is a form of social recommendation but there are many “social links” that remain absent from Google’s search algorithm. Facebook, del.icio.us, Twitter, and many other sites all have links that are absent from Google’s search algorithm. Theoretically these links should be considered when calculating the value of a page but unfortunately Google doesn’t have access to that information.
Developing a Social Search Algorithm
One could argue that the links shared through social platforms can be compiled to create a search engine just as powerful if not more powerful than Google. When I go searching for Wordpress templates currently, I look through del.icio.us not Google because I know that the del.icio.us community is more effective at finding web development related items.
Facebook currently has over 20 million pieces of content that are “shared” each month according to the site’s statistics page. This adds up to a lot of information that can be used for developing a more effective search algorithm. Facebook also has the ability to include a number of other variables into their search algorithm including:
- Comments on shared stories
- “Friend Rank” - A way of valuing the relevance of a link based on a user’s authority via the number of friends they have and the influence of each of those friends
- Number of times a story is shared
- Networks users are members of (good for targeted search)
- … and more
All of these variables are effective for developing a search engine that is just as powerful if not more powerful than Google. Currently Facebook has temporarily granted Microsoft the ability to display Live Search within their site in exchange for guaranteed yearly ad revenue. I’d also imagine that Microsoft’s sizeable investment in Facebook that valued the company at around $15 billion was also tied to a search agreement.
I believe that agreement provides only a temporary revenue source for Facebook while they work toward developing a much more effective search algorithm.
Google’s Inroads
Google has been trying to make inroads in social search. They recently launched “SearchWiki” which lets users comment and vote on search results. They hope that this voting system will help improve the search results in the long run. Google may have something going there but personally I don’t think this is the most effective way of sorting search results.
The problem is that the results have already been sorted prior to users voting on them. If someone said “Here are ten articles, tell us which article is most relevant to this search”, it would be a great polling model but what if the most relevant article isn’t among those ten options? Google could argue that their search system is already effective at finding the top 10, and their position would be a strong one given that they’re the largest search engine.
Others Make Attempts
Numerous other companies are also participating in the social search race. Mahalo has been trying to create a social search engine which is based on information submitted by their users. Digg lets users vote on their stories, but Digg is also known for its historically biased community. As I previously mentioned, del.icio.us has a great search tool as well. Twitter could also take a shot at the social search race.
All of these sites have interesting models but as Aditya Agarwal mentioned, nobody has a most accurate image of the social graph. In the world of social search, having the most accurate social graph is arguably as important as it is for modern search engines (Google, Yahoo, Live, etc) to have the most robust index of content on the web.
Conclusion
For the time being social search is something that no company has effectively tackled. I’d argue that Facebook is in the position to be the most accurate social search engine in the world. The question is not “if” they’ll launch an effective search engine but when. Are there any other companies that you think are properly positioned for becoming a leading social search engine?
Hey Nick, good article (as always). It’s also worth considering that if you filter search results by your explicit social graph - and your particular social graph is not large - the results are likely to be as biased as some of the other community-based services that you mention. Full disclosure… i work at http://www.oneriot.com and we’re tackling this problem head-on - ranking search results based on what millions of people are talking about right now for any subject. For example, a search for “apple” on OneRiot today will bring up stories about Steve Job’s health (i.e. what people are talking about right now) http://www.oneriot.com/search?q=apple vs the navigational results you’ll get with Google, Live, etc http://search.live.com/results.aspx?q=apple&form=QBLH As a cute aside, we also have Facebook Connect integration - so when a user finds a great story or video on OneRiot, they can easily share it with their Facebook friends. (sorry, this is becoming a “pitch” now, not a blog comment… i’ll stop
That was a really interesting post and a really good point. But, the comment above mine also was interesting point and the reason why it will probably be a long time in the making. Thanks for another good one!
I think Facebook has some really great plans tucked away that they are just waiting for the right time to roll out.
Here is what I think: as the Internet becomes too big and search becomes impersonal and brand mkt-$$ oriented, you start to have diminishing returns for the end user. It is like when you live in a small city and you have a few merchants on the city catalog, so it is great you can go to one place to find their phone numbers and pretty much you know the ones that are good vs. the ones that suck (many times through the back channels of your social network). Compare that to the thick yellow pages of a big city where you don’t know which of those merchants are any good, and many times just go for the known brand or bigger ad (again $$ dollars), sometimes with disappointing results.
Now FB Has the opportunity to enable the “small town trust links” on a much bigger scale across all the info it gathers through your profile and friends. The key piece that it needed was FB Connect, so that info from the outer internet flows back to FB. If (or when) Connect reaches critical mass, you will be able to quickly find answers to the most common and profitable queries (e.g. queries that lead to a high value transaction) from trusted sources in FB (your friends).
In this new high-trust, personal world, add placement becomes less relevant (Google better execute on non-search stuff). My favorite restaurants almost never advertise, they can just rely on world of mouth.
This is a awesome article. i like the point the comment above me make about the search queries that will lead to high value transaction. Digital word of mouth can add credibility to the smallest of brands. Google’s algorithm is based on publishers linking to other publishers. Facebook would be based on friends sharing content which in the end would more credible and satisfying for the user. Great Potential if executed properly.